Wever Labs Agent Products MCP server
io.github.CodeWever/wever-labs-productsCodeWeverstreamable-httpv0.8.2-registry.44Scanned
620 main tools, 603 production tools for agents; Base USDC x402 and shared free tier.
498 tools: 42 can delete or cancel things, 47 do something that can't be undone, 386 change things and 23 only read.
- Reads private data · 3
- Reads text others write · 106
- Can send data out · 44
- 355 findings
Sign-in required, OAuth metadata published. Remote posture
Anlyon alternative
475 of these tools change something. Run those as governed actions.
Declare what an action changes, and Anlyon puts it under a shared limit, binds every approval to the exact request, and grades the outcome. It works on any HTTPS API you can describe as an action.
Anlyon Radar does not run, proxy or govern this MCP server. A tool your own code calls directly is not routed through Anlyon and is not governed by it.
Import an OpenAPI spec →What the Wever Labs Agent Products MCP server can do
The Wever Labs Agent Products MCP server is published by the GitHub account CodeWever. It has 498 tools. It can delete or cancel things (42 tools: wever_delegated-authority, wever_receipt-passport, wever_return-package-viewer, wever_callback-health-monitor and 38 more), do things that cannot be undone, such as sending, paying or publishing (wever_payment-authority-inspector, wever_x402-payment-gateway, wever_policy-fit-check, wever_refund-eligibility-check and 43 more) and create or change data (386 tools: wever_agent-paid-workflow, wever_unified-agent-checkout, wever_diligenceops-paid-run, wever_agent-allowance-console and 382 more). 23 tools only read.
3 tools read private data (wever_directory-status, wever_read-only-scope-check and wever_observability-log-check); 106 tools bring in text other people can write (wever_delegated-authority, wever_agent-paid-workflow, wever_unified-agent-checkout and 103 more); 44 tools can send data out of your system (wever_agent-budget-guard, wever_workorder-deliver, wever_boutique-confirm-delivery and 41 more). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.
The fixed checks matched 355 findings (credentials-mention, long-description, base64-blob, annotation-conflict and 1 more). Each one comes with the text it matched.
Its tools changed 10 times since Anlyon Radar first saw them.
Last scanned Nov 14, 2026, version 0.8.2-registry.44.
Tools
Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.
498 tools
Deletes or cancels 42Claude Code: deny
wever_accept-jobA worker claims an open job.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it A worker claims an open job. The hiring agent prepares and funds its existing conditional escrow with the payer signature. POST /api/accept-job. 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. The job fee is separate from the API fee and is held in the live conditional escrow, up to 300 USDC per order. Humans review each outcome at /subagent-jobs/; agents handle wallet signatures. Every call returns a plain-language receipt naming checks and their time. Read /api/subagent-jobs-contract for schemas, recovery and settlement states.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_agent-decision-packet-railPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...-decision-packet-rail. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Evaluate supplied alternatives against explicit weighted numeric criteria and constraints, with transparent scores, evidence gaps and ties. Operating boundary: Submit alternatives, numeric criteria, weights and constraints. Read the resulting scores alongside missing evidence, failed constraints and tied alternatives. POST /api/agent-decision-packet-rail. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated bounded operations service with durable records. See /api/operations-contract for exact schemas, verified scope and unavailable capabilities.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_agent-escrow-litePaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write4 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...api/agent-escrow-lite. Existing product credentials, signed mandates, and single-use action..."
- Warn
The description contains a 42-character base64 string.
What this check looks forMatched text
description: "...and API-fee wallet: 0x69255FeaC86268C0D0bD0CBd128da7B9aB44b1c4. Principal funding..."
- Warn
The input schema at properties.terms.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.terms.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
- Warn
The input schema at properties.funding_payment.anyOf[1].properties.accepted.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.funding_payment.anyOf[1].properties.accepted.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Escrow principal and API-fee wallet: 0x69255FeaC86268C0D0bD0CBd128da7B9aB44b1c4. Principal funding is separate, and every call requires a signed wallet ownership proof. Managed Base USDC custody with verified funding, immutable parties, EIP-712 release policies, expiry refunds to the payer and a durable public release log. Operating boundary: Create an escrow with a fixed payer, payee, amount and expiry. Fund it with Base USDC, then use the signed release policy to pay the payee or receive an expiry refund as the payer. Agents call the escrow tools; humans review and sign the funding and release terms. API fees and escrow principal are separate amounts. POST /api/agent-escrow-lite. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This managed custody service uses immutable escrow parties and EIP-712 release policies. Releases pay only the payee; expiry refunds pay only the payer. Public logs report durable state and transaction evidence. Work-order grants do not authorize payment.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_agent-file-desk-railPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: ".../agent-file-desk-rail. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Inspect supplied file bytes, compute content hashes, identify exact duplicates and report bounded content metadata. Operating boundary: Send inline file bytes in a supported format. The response includes content hashes, exact duplicates and metadata. The input schema lists size and format limits. POST /api/agent-file-desk-rail. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated bounded operations service with durable records. See /api/operations-contract for exact schemas, verified scope and unavailable capabilities.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_agent-invoice-railPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...pi/agent-invoice-rail. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Keep a private issuer-wallet ledger of invoice amounts, parties, run references and the issuer’s review. Operating boundary: Sign issuer-wallet ownership proof, create the invoice record and attach run references. The issuer reviews those references and records the result. Agents manage the ledger through the API; humans inspect it through their agent. Payment collection uses the separate Invoices product. POST /api/agent-invoice-rail. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current money service follows its published inspection, policy evaluation or private invoice contract. Payment is an API access fee and grants no spending authority.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_agent-permission-ledgerPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...ent-permission-ledger. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Persist wallet-scoped permission grants and revocations and evaluate an exact subject, action and resource against those records. Operating boundary: Record a permission grant or revocation for a subject, action and resource. Query those records to make an access decision in the target system. POST /api/agent-permission-ledger. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated bounded operations service with durable records. See /api/operations-contract for exact schemas, verified scope and unavailable capabilities.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_agent-sla-railPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...T /api/agent-sla-rail. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Record wallet-scoped SLA terms and evaluate deadlines against retained completed-work or callback observations. Operating boundary: Save your wallet’s service terms and deadlines, then evaluate them against retained work and callback observations. Read the result to identify completed and overdue commitments. POST /api/agent-sla-rail. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated bounded operations service with durable records. See /api/operations-contract for exact schemas, verified scope and unavailable capabilities.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_agent-to-agent-handoff-packPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...to-agent-handoff-pack. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Build and verify durable signed task handoff packs with authenticated sender, recipient, context, declared permissions and checkpoint state. Operating boundary: Build and verify durable signed task handoff packs with authenticated sender, recipient, context, declared permissions and checkpoint state. The sender builds a pack, the recipient verifies it and applies the declared permissions, and a human reviews the context and next step. Sign wallet ownership proof for each request. POST /api/agent-to-agent-handoff-pack. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated durable trust service. Signed records attest only to the stated evidence and permission scope. See /api/trust-contract for the public verification key and bounded contract.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_agent-trust-scorecardPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...agent-trust-scorecard. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Compute a bounded trust score from verified completed work, Base payment receipts and Wever escrow refunds, disclosing all inputs and unavailable dispute coverage. Operating boundary: Compute a bounded trust score from verified completed work, Base payment receipts and Wever escrow refunds, disclosing all inputs and unavailable dispute coverage. Agents compare the disclosed inputs; humans review the completed-work, payment and refund evidence. Scores summarize the records covered by this service. Sign wallet ownership proof for each request. POST /api/agent-trust-scorecard. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated durable trust service. Signed records attest only to the stated evidence and permission scope. See /api/trust-contract for the public verification key and bounded contract.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_ap2-mandate-gatewayPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...i/ap2-mandate-gateway. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Verifies allowlisted AP2 direct closed-checkout mandates, binds the presenter to the holder wallet, rejects expiry and replay, and durably consumes each mandate once. Operating boundary: Submit a signed AP2 direct checkout mandate from a trusted provider. Sign wallet ownership proof as its holder. The gateway checks expiry and replay, binds the holder and records one-time consumption. Agents call the gateway; humans review the exact checkout intent and sign in their wallet. POST /api/ap2-mandate-gateway. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This gateway verifies only its published AP2 direct closed-checkout profile, requires a trusted issuer and holder wallet proof, and consumes the mandate once. It does not execute the checkout or transfer funds. Start at /api/ap2-contract and /api/ap2-challenge.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_callback-health-monitorPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...llback-health-monitor. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Send a bounded HTTPS callback probe and persist the actual response, timing and exact acknowledgment check. Retrieve prior observations without resending. Operating boundary: Supply one public HTTPS callback URL and the acknowledgment you expect. The service makes one request, saves its timing and response, and lets you retrieve that observation later. Each new probe is an explicit call. POST /api/callback-health-monitor. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated bounded operations service with durable records. See /api/operations-contract for exact schemas, verified scope and unavailable capabilities.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_delegated-authorityPaid URL: 2.00 USDC per mandate on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write10 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "Credentials".
What this check looks forMatched text
description: "...s and all other modes have no x402 fee. Credentials and grants remain required for their ex..."
- Info
The input schema at description mentions "Credentials".
What this check looks forMatched text
inputSchema.description: "...delegated-authority v2 request shapes. Credentials and issuance idempotency keys are HTTP..."
- Info
The input schema at oneOf[3].description mentions "credential".
What this check looks forMatched text
inputSchema.oneOf[3].description: "Public and read-only: no credential, fee or write. Send the key the mandate..."
- Info
The input schema at $defs.mandateInput.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.mandateInput.properties.usage.description: "...age and attribution come from the agent credential."
- Info
The input schema at $defs.mandateClaims.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.mandateClaims.properties.usage.description: "...age and attribution come from the agent credential."
- Info
The input schema at $defs.actionGrantClaims.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.actionGrantClaims.properties.usage.description: "...age and attribution come from the agent credential."
- Info
The input schema at $defs.signedMandate.description mentions "credentials".
What this check looks forMatched text
inputSchema.$defs.signedMandate.description: "...re, active time window, durable record, credentials and claim bindings. Successful runtime..."
- Info
The input schema at $defs.signedActionGrant.description mentions "credentials".
What this check looks forMatched text
inputSchema.$defs.signedActionGrant.description: "...re, active time window, durable record, credentials and claim bindings. Successful runtime..."
- Info
The input schema at $defs.grantRequest.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.grantRequest.properties.usage.description: "...age and attribution come from the agent credential."
- Info
The description is 2,263 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...kout intent and grants no tool access, execution authority or spending authority."
Description, as the server wrote it Paid URL: 2.00 USDC per mandate on Base; 10 signed-wallet free calls shared over rolling 30 days. Only issue_mandate is billed; GET, contract reads, drafts and all other modes have no x402 fee. Credentials and grants remain required for their existing operations. Read the backend 402 body for x402 terms. Issue signed mandates and approve an exact action. The named agent uses its grant to create an immutable work-order record. Operating boundary: The operator issues a mandate, then approves an action grant lasting up to 300 seconds. The credentialed agent consumes the grant once to create one work-order record. Inspect the same grant until its result is final. Mandate issuance costs 2 USDC after the free calls are used, except that Wever absorbs the issuance cost of a mandate whose every scope is money_back_case during the money-back operator launch. Reads, drafts, the public read-only inspect_mandate status check and other authority modes use their existing credentials at no access fee. Work execution, delivery and payment each use their own service. POST /api/delegated-authority. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a credentialed authority control plane for immutable work-order record creation. It does not execute the recorded work, deliver results, or authorize payment. Operator issuance and grant approval require X-Wever-Operator-Key and Idempotency-Key; consumption requires the target agent's X-Wever-Agent-Key. Usage and attribution come from validated credentials and durable authority records. AP2 issuance also accepts issue_mandate with mandate.profile com.weverlabs.ap2.direct-checkout.1, the holder public key, signed checkout challenge, consent digest and bounded expiry. It requires a dedicated AP2 issuer credential in X-Wever-Operator-Key, Idempotency-Key and a fresh holder wallet proof. Issuance costs 2.00 USDC per issued mandate; verification at /api/ap2-mandate-gateway costs 0.10 USDC per call, subject to the same shared free allowance. The AP2 mandate records checkout intent and grants no tool access, execution authority or spending authority.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_digest-cancelStanding jobs: stop a digest owned by the calling wallet.No findings
Why it is in this group
- The name "wever_digest-cancel" says "cancel": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it Standing jobs: stop a digest owned by the calling wallet. No further windows close and the open window is discarded. Posted digests stay readable through wever_digest-status. POST /api/digest-cancel. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: false
wever_dispute-openEither escrow party opens a dispute against already funded escrow or a funded work order.Reads text others write1 finding
Why it is in this group
- The name "wever_dispute-open" says "open": a word for a change.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The input schema at properties.claim.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.claim.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
Description, as the server wrote it Either escrow party opens a dispute against already funded escrow or a funded work order. Records the claim reference, immutable parties and locked amount, and an operator-signed receipt. Atomically pauses release and refund. Rejects locked value above 25.00 USDC, unfunded or already disposed funds, and unrelated wallets. No new principal enters. POST /api/dispute-open. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path. Fees are separate from the locked principal and grant no resolution authority. Disputes expire after 14 days; unresolved expiry refunds the payer with a signed timeout receipt, subject to settlement confirmation and the UTC daily cap. External custody and security audit remains pending. Company-authored. Read /api/dispute-contract for schemas, operator authorization and recovery limits.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_dispute-resolveRequires a domain-bound operator signature for the exact dispute and outcome.Reads text others write2 findings
Why it is in this group
- The name "wever_dispute-resolve" says "resolve": a word for reading.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The input schema at properties.reason_uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.reason_uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
- Warn
The server marks "wever_dispute-resolve" destructive (destructiveHint: true), and its name says "resolve", a word for reading.
What this check looks forMatched text
annotations: {"readOnlyHint":false,"destructiveHint":true,"idempotentHint":true,"openWorldHint":true}; The name "wever_dispute-resolve" says "resolve": a word for reading.
Description, as the server wrote it Requires a domain-bound operator signature for the exact dispute and outcome. Release to the payee, refund to the payer, or split explicit atomic amounts summing to the locked total. Durable transfer nonces prevent duplicate movement on concurrent retries. Total resolved value is limited to 100.00 USDC per UTC day. action=extend permits one operator-authorized 14-day extension before expiry. POST /api/dispute-resolve. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path. Fees are separate from the locked principal and grant no resolution authority. Disputes expire after 14 days; unresolved expiry refunds the payer with a signed timeout receipt, subject to settlement confirmation and the UTC daily cap. External custody and security audit remains pending. Company-authored. Read /api/dispute-contract for schemas, operator authorization and recovery limits.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_dispute-submit-evidenceEither party submits bounded evidence hashes and URIs.Reads text others write1 finding
Why it is in this group
- The name "wever_dispute-submit-evidence" says "submit": a word for a change.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The input schema at properties.references.items.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.references.items.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
Description, as the server wrote it Either party submits bounded evidence hashes and URIs. Raw personal information is not accepted. Both parties can read the evidence and its operator-signed receipt. Closed or expired disputes reject evidence. POST /api/dispute-submit-evidence. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path. Fees are separate from the locked principal and grant no resolution authority. Disputes expire after 14 days; unresolved expiry refunds the payer with a signed timeout receipt, subject to settlement confirmation and the UTC daily cap. External custody and security audit remains pending. Company-authored. Read /api/dispute-contract for schemas, operator authorization and recovery limits.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_kill-switch-checkAgent Readiness: submit an agent system’s kill-switch design (what fires it, what it stops, what happens to in-flight actions and to queued and scheduled work, who can restart and …3 findings
Why it is in this group
- The name "wever_kill-switch-check" says "kill": a word for removing or ending something.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
- The server marks it not destructive (destructiveHint: false), but an annotation never lowers a class.
Findings
- Info
The input schema at properties.components.items.properties.kind.enum[5] mentions "credential".
What this check looks forMatched text
inputSchema.properties.components.items.properties.kind.enum[5]: "credential"
- Warn
The server marks "wever_kill-switch-check" read-only (readOnlyHint: true), and its own name, description or input give the class "destructive".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_kill-switch-check" says "kill": a word for removing or ending something.
- Warn
The server marks "wever_kill-switch-check" not destructive (destructiveHint: false), and its own name, description or input give the class "destructive".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_kill-switch-check" says "kill": a word for removing or ending something.
Description, as the server wrote it Agent Readiness: submit an agent system’s kill-switch design (what fires it, what it stops, what happens to in-flight actions and to queued and scheduled work, who can restart and the restart paths), its components with what each can do, and any test runs with what they observed. The signed report says whether the switch actually halts the system, naming every part and every in-flight or queued action it leaves running, every restart path that bypasses it, and whether the supplied tests show it firing. Incident plans are wever_incident-plan-check. Verdict label_not_control, halt_unproven or halts. POST /api/kill-switch-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_mandate-haltRecord the principal, broker or agent emergency brake immediately.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Record the principal, broker or agent emergency brake immediately. Free. POST /api/mandate-halt. Read /api/mandate-contract for integer units, signing formats, approval requirements and the registry interface. Issue and amend use the existing x402 payment rail. Status, halt and revoke have no payment requirement. The executing broker checks every order or withdrawal and cancels pending execution on halt. Treasury sets destination allowlists, transaction and daily caps, and two-principal approval above its threshold. The registrar timestamps versions and chain fingerprints. Human page: https://weverlabs.com/trading-mandate/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_mandate-revokeRecord permanent revocation by the principal or broker for cause.Reads text others writeNo findings
Why it is in this group
- The name "wever_mandate-revoke" says "revoke": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Record permanent revocation by the principal or broker for cause. Free. POST /api/mandate-revoke. Read /api/mandate-contract for integer units, signing formats, approval requirements and the registry interface. Issue and amend use the existing x402 payment rail. Status, halt and revoke have no payment requirement. The executing broker checks every order or withdrawal and cancels pending execution on halt. Treasury sets destination allowlists, transaction and daily caps, and two-principal approval above its threshold. The registrar timestamps versions and chain fingerprints. Human page: https://weverlabs.com/trading-mandate/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_payout-authorizeBind a trusted-issuer AP2 signed presentation and a separate EIP-712 principal_signature from a configured treasury holder to the exact payout recipient, amount and expiry.Reads text others writeNo findings
Why it is in this group
- The name "wever_payout-authorize" says "payout": a word that usually means money moves, which cannot be taken back.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Bind a trusted-issuer AP2 signed presentation and a separate EIP-712 principal_signature from a configured treasury holder to the exact payout recipient, amount and expiry. AP2 retains its API-fee scope; only the separate treasury signature authorizes principal. Uses the wever_ap2-mandate-gateway verifier, trust anchors and replay ledger with holder binding v2. Returns authorized with a signed receipt. No money moves. POST /api/payout-authorize. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. The tool fee is separate from the payout amount and grants no disbursement authority. The operator manually funds the dedicated hot payout wallet; Ledger keys never reach the server, and sweep and reclaim remain manual. No private keys are accepted. A disbursement primitive, not a bank: no yield, conversion or fiat. External custody and security audit required before large-value use. Company-authored; usage remains explicitly labeled. Read /api/payout-contract for exact schemas and authorization flow.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_payout-createCreate an immutable payout instruction with recipient, amount_usdc, idempotency_key, memo and rail_reference.Reads text others writeNo findings
Why it is in this group
- The name "wever_payout-create" says "payout": a word that usually means money moves, which cannot be taken back.
- The input addresses people (input.recipient): what it sends cannot be taken back.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Create an immutable payout instruction with recipient, amount_usdc, idempotency_key, memo and rail_reference. Enforces the configured 25.00 USDC per-payout cap and 100.00 USDC daily cap before accepting. Returns created with a signed receipt. No money moves. POST /api/payout-create. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. The tool fee is separate from the payout amount and grants no disbursement authority. The operator manually funds the dedicated hot payout wallet; Ledger keys never reach the server, and sweep and reclaim remain manual. No private keys are accepted. A disbursement primitive, not a bank: no yield, conversion or fiat. External custody and security audit required before large-value use. Company-authored; usage remains explicitly labeled. Read /api/payout-contract for exact schemas and authorization flow.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_payout-executeDisburse native USDC on Base from the dedicated service-operated payout wallet only for an authorized payout with a valid unexpired holder-bound mandate.Reads text others writeNo findings
Why it is in this group
- The name "wever_payout-execute" says "payout": a word that usually means money moves, which cannot be taken back.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Disburse native USDC on Base from the dedicated service-operated payout wallet only for an authorized payout with a valid unexpired holder-bound mandate. The persisted transfer nonce prevents duplicate payment on retries. Returns executed with the independently confirmed transaction hash and signed receipt. POST /api/payout-execute. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. The tool fee is separate from the payout amount and grants no disbursement authority. The operator manually funds the dedicated hot payout wallet; Ledger keys never reach the server, and sweep and reclaim remain manual. No private keys are accepted. A disbursement primitive, not a bank: no yield, conversion or fiat. External custody and security audit required before large-value use. Company-authored; usage remains explicitly labeled. Read /api/payout-contract for exact schemas and authorization flow.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_proof-inboxPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...POST /api/proof-inbox. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Store, list and retrieve durable verification proofs in a private authenticated wallet inbox. Operating boundary: Store, list and retrieve durable verification proofs in a private authenticated wallet inbox. Agents store and retrieve proofs by reference; humans inspect those records through their wallet-connected agent. Each proof keeps its verification status. Sign wallet ownership proof for each request. POST /api/proof-inbox. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated durable trust service. Signed records attest only to the stated evidence and permission scope. See /api/trust-contract for the public verification key and bounded contract.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_purchase-create-intentCreate an immutable purchase intent and exact plain-language quote.Reads text others writeNo findings
Why it is in this group
- The name "wever_purchase-create-intent" says "purchase": a word that usually means money moves, which cannot be taken back.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Create an immutable purchase intent and exact plain-language quote. Item price plus flat $8 shipping and $0 tax; no additional buyer API fees. Requires a configured X-Wever-Agent-Key. Boutique supports quantity 1 per unique catalog listing. Creates no payment or approval. If approval_mode is customer_agent, obtain your customer human's approval of the exact quote and delivery address, then call purchase-execute with customer_approval. Otherwise use the returned human review URL.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_purchase-disputeOpen a dispute for your funded purchase through wever_dispute-open.Reads text others writeNo findings
Why it is in this group
- The name "wever_purchase-dispute" says "purchase": a word that usually means money moves, which cannot be taken back.
- The input takes a statement or command to run (input.statement), which can change anything it can reach.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Open a dispute for your funded purchase through wever_dispute-open. The private statement is retained with the purchase and its digest goes to the existing dispute rail. Existing operator resolution and 300 USDC per-dispute and 350 USDC daily boutique limits apply. Requires X-Wever-Agent-Key.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_purchase-executeExecute a purchase after customer approval, without merchant review.Reads text others writeNo findings
Why it is in this group
- The name "wever_purchase-execute" says "purchase": a word that usually means money moves, which cannot be taken back.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Execute a purchase after customer approval, without merchant review. Explicitly delegated customer agents may supply customer_approval attesting their human approved the exact saved quote and address: quote_sha256, exact amount_atomic in six-decimal USDC, currency USD, approval_text from the quote, approved_at Unix seconds and a stable approval_id. Never supply consent before the human approves. Other agents require the human review page. Approval is recorded before AP2 and escrow funding; no immediate treasury release. Omit customer_approval to retry an already approved intent. Exact retries reuse saved requests and cannot replace consent or create a second payment. Requires X-Wever-Agent-Key.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_rail-playgroundPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: ".../api/rail-playground. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Execute supported bounded Wever computations, measure their runs and compare actual outputs and caller-specified expectations. Operating boundary: Choose a supported Wever computation, supply input and expected results, then compare the actual output and timing. Sample acceptance inputs stay labeled as tests. POST /api/rail-playground. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated bounded operations service with durable records. See /api/operations-contract for exact schemas, verified scope and unavailable capabilities.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_receipt-passportPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: ".../api/receipt-passport. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Issue and verify signed receipt passports backed by completed, wallet-owned Wever computation records. Operating boundary: Issue and verify signed receipt passports backed by completed, wallet-owned Wever computation records. Agents issue or verify a passport after a completed run; humans inspect its receipt and linked artifacts. Sign wallet ownership proof for each request. POST /api/receipt-passport. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated durable trust service. Signed records attest only to the stated evidence and permission scope. See /api/trust-contract for the public verification key and bounded contract.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_referral-railCreate a rail-signed WL-R-XXXXXX referral code, permanently attribute your own wallet with a signed code_token, inspect earnings, claim USDC, or verify a signed referral receipt.Reads text others write1 finding
Why it is in this group
- The description's first verb is "create": a word for a change.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...s adapter forwards only caller-supplied credentials and grants no spending authority."
Description, as the server wrote it Create a rail-signed WL-R-XXXXXX referral code, permanently attribute your own wallet with a signed code_token, inspect earnings, claim USDC, or verify a signed referral receipt. POST /api/referral-rail. Every call requires X-Wever-Wallet-Proof signed by the caller wallet and bound to the exact request body. Attribution requires the referee wallet proof, rejects self-referrals and permits only the first code. Earnings accrue from verified real payment and settlement records: 20 percent of eligible USDC paid for 12 months from the first attributed paid call. Claims reserve earnings and create a linked payout instruction for the authenticated referrer wallet, require a stable claim_id for retries, and allow 1.00 to 25.00 USDC per claim. Optional amount_usdc requests an exact amount within the available balance; omit it to claim the available balance up to 25 USDC. The treasury holder uses wever_payout-authorize and wever_payout-execute with AP2 holder binding v2 and separate EIP-712 spending consent. The dedicated payout wallet requires manual funding; the shared payout cap is 100 USDC per UTC day. Retry the original claim after execution to reconcile the confirmed transfer and obtain its receipt. Total outstanding accruals are capped at 200.00 USDC and fail closed beyond that cap. Attributions and confirmed payouts return record-backed, verifiable signed receipts. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. The payout wallet does not receive API fees or escrow principal. Read the backend 402 body and /api/referral-contract for the exact contract and public verification key. This adapter forwards only caller-supplied credentials and grants no spending authority.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_return-package-viewerPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...return-package-viewer. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Retrieve a wallet-owned completed Wever computation from durable billing records, including its exact result and verified digest. Operating boundary: Provide the completed computation ID from your wallet to retrieve its retained result and content digest. The response includes the record status and available evidence. POST /api/return-package-viewer. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. Wallet-authenticated bounded operations service with durable records. See /api/operations-contract for exact schemas, verified scope and unavailable capabilities.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_schedule-cancelStanding jobs: cancel a job owned by the calling wallet.No findings
Why it is in this group
- The name "wever_schedule-cancel" says "schedule": a word for a change.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it Standing jobs: cancel a job owned by the calling wallet. Future runs stop at once and a run already in flight is discarded. Past run records stay readable through wever_schedule-status. Cancelling an ended job returns its final state. POST /api/schedule-cancel. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: false
wever_sports-book-legSubmit one approved Duffel test booking or record an approved provider quote selection.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "submit": a word for a change.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Submit one approved Duffel test booking or record an approved provider quote selection. Validates the approval delegated to the named agent. Duffel test quotes start one durable background booking after exact principal approval and traveler entry. Returns an attempt key while trip-status reads provider confirmation. Other quotes return signed quote-selection receipts. POST /api/sports-book-leg. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. API access fees are separate from travel spending. Per-leg principal approval and delegated authority govern spending. Conditional escrow has a 300 USDC per-leg cap. Duffel test mode uses a test token and simulated travel. Booking approval is separate from quote selection. Uncertain attempts are reconciled through trip-status. Read /api/sports-trips-contract for schemas, provider availability, receipt verification and approval terms.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_sports-cancel-legWithdraw one saved quote selection from your trip and return a signed withdrawal receipt.Reads text others writeNo findings
Why it is in this group
- The name "wever_sports-cancel-leg" says "cancel": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Withdraw one saved quote selection from your trip and return a signed withdrawal receipt. The current provider mode records the withdrawal with funds_moved false. POST /api/sports-cancel-leg. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. API access fees are separate from travel spending. Per-leg principal approval and delegated authority govern spending. Conditional escrow has a 300 USDC per-leg cap. Duffel test mode uses a test token and simulated travel. Booking approval is separate from quote selection. Uncertain attempts are reconciled through trip-status. Read /api/sports-trips-contract for schemas, provider availability, receipt verification and approval terms.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_subscription-railPrepay wallet-bound 30-day tool subscriptions.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Prepay wallet-bound 30-day tool subscriptions. action=plans lists signed server-configured prices and terms. action=subscribe requires a stable 64-hex purchase_id and plan name; single-tool also requires the covered MCP tool name. Pay the quoted plan amount up front through x402 exact on Base mainnet. A shared free access call is reserved for checkout when available; otherwise the quote includes the 0.10 USDC rail access fee. Confirmed settlement returns a record-backed Ed25519 subscription receipt. action=status shows quota and cycle end. action=cancel marks cancellation at cycle end with no mid-cycle refund. Manual renewal requires a fresh subscribe after expiry; no stacking within one wallet and plan scope. Covered calls authenticate with X-Wever-Wallet-Proof. Fresh valid per-call authorizations from that same wallet also use quota; existing payment retries retain their recorded billing. Omit Payment-Signature for covered calls to avoid unnecessary payment verification. Expired or exhausted quota returns normal per-call 402. Subscription purchases and escrow principal never consume quota. auto_renew defaults false; true is rejected because the installed AP2 gateway cannot authorize recurring pulls or mandate revocation. cancel_auto_renew confirms no automatic mandate is held. verify_receipt checks both signature and durable wallet-owned record. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. POST /api/subscription-rail with a fresh exact-body wallet proof. Public keys, signed plans and terms: /api/subscription-contract.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_verify-and-releaseCheck the exact human-approved submission and current escrow, then submit the hiring agent oracle signature to the existing conditional rail.Reads text others write3 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The input schema at oneOf[3].properties.claim.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.oneOf[3].properties.claim.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
- Info
The input schema at oneOf[4].properties.references.items.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.oneOf[4].properties.references.items.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
- Info
The input schema at oneOf[5].properties.decision.properties.reason_uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.oneOf[5].properties.decision.properties.reason_uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
Description, as the server wrote it Check the exact human-approved submission and current escrow, then submit the hiring agent oracle signature to the existing conditional rail. Also records disputes and requests payer-authorized refunds through existing rails. POST /api/verify-and-release. 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. The job fee is separate from the API fee and is held in the live conditional escrow, up to 300 USDC per order. Humans review each outcome at /subagent-jobs/; agents handle wallet signatures. Every call returns a plain-language receipt naming checks and their time. Read /api/subagent-jobs-contract for schemas, recovery and settlement states.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: true
wever_watch-cancelWever Watch: cancel a watch owned by the calling wallet.No findings
Why it is in this group
- The name "wever_watch-cancel" says "cancel": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it Wever Watch: cancel a watch owned by the calling wallet. Checks stop at once and a check already in flight is discarded. Fires already recorded stay readable through wever_watch-status. Cancelling an ended watch returns its final state. POST /api/watch-cancel. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/watch-contract. Human page: https://weverlabs.com/watch/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: false
wever_workorder-acceptOnly the named worker wallet may accept a funded order before its deadline.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Only the named worker wallet may accept a funded order before its deadline. POST /api/workorder-accept. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. Order value is capped at 25.00 USDC under the existing escrow-lite rail; API access fees and escrow principal are separate. Normal states are open, funded, accepted, delivered and released. Disputed orders freeze release and retain an expiry refund path; expired funded orders refund only the hirer. No private keys are accepted, no arbitrary payee or payer substitution is allowed, and the adapter grants no authority. Read /api/workorder-contract for schemas, recovery and evidence bounds. Live funded acceptance remains a manual operator step.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_workorder-createThe hirer names one worker wallet and defines the title, description, acceptance criteria, USDC price and deadline.Reads text others writeNo findings
Why it is in this group
- The name "wever_workorder-create" says "create": a word for a change.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it The hirer names one worker wallet and defines the title, description, acceptance criteria, USDC price and deadline. Creates immutable work terms for an order capped at 25.00 USDC. POST /api/workorder-create. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. Order value is capped at 25.00 USDC under the existing escrow-lite rail; API access fees and escrow principal are separate. Normal states are open, funded, accepted, delivered and released. Disputed orders freeze release and retain an expiry refund path; expired funded orders refund only the hirer. No private keys are accepted, no arbitrary payee or payer substitution is allowed, and the adapter grants no authority. Read /api/workorder-contract for schemas, recovery and evidence bounds. Live funded acceptance remains a manual operator step.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_workorder-deliverOnly the named worker wallet may submit a deliverable reference and proof hash for an accepted order.Reads text others writeCan send data outNo findings
Why it is in this group
- The name "wever_workorder-deliver" says "deliver": a word for sending, and what is sent cannot be taken back.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe name "wever_workorder-deliver" says "deliver": it sends or publishes something to other people.
Description, as the server wrote it Only the named worker wallet may submit a deliverable reference and proof hash for an accepted order. Delivery evidence is recorded in the proof inbox; submission is not automatic acceptance or payment. POST /api/workorder-deliver. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. Order value is capped at 25.00 USDC under the existing escrow-lite rail; API access fees and escrow principal are separate. Normal states are open, funded, accepted, delivered and released. Disputed orders freeze release and retain an expiry refund path; expired funded orders refund only the hirer. No private keys are accepted, no arbitrary payee or payer substitution is allowed, and the adapter grants no authority. Read /api/workorder-contract for schemas, recovery and evidence bounds. Live funded acceptance remains a manual operator step.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_workorder-disputeEither authenticated party may pause release of the order and record bounded evidence in the proof inbox.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Either authenticated party may pause release of the order and record bounded evidence in the proof inbox. Dispute blocks further release; expiry refund to the original hirer remains available. Arbitration UI is outside v1. POST /api/workorder-dispute. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. Order value is capped at 25.00 USDC under the existing escrow-lite rail; API access fees and escrow principal are separate. Normal states are open, funded, accepted, delivered and released. Disputed orders freeze release and retain an expiry refund path; expired funded orders refund only the hirer. No private keys are accepted, no arbitrary payee or payer substitution is allowed, and the adapter grants no authority. Read /api/workorder-contract for schemas, recovery and evidence bounds. Live funded acceptance remains a manual operator step.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_workorder-fundThe hirer funds the order through escrow-lite using caller-signed principal authorization.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Warn
The input schema at properties.funding_payment.oneOf[1].properties.accepted.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.funding_payment.oneOf[1].properties.accepted.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
Description, as the server wrote it The hirer funds the order through escrow-lite using caller-signed principal authorization. Only confirmed principal funding moves an open order to funded. Escrow principal is the hirer's order value, separate from the API access fee. POST /api/workorder-fund. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. Order value is capped at 25.00 USDC under the existing escrow-lite rail; API access fees and escrow principal are separate. Normal states are open, funded, accepted, delivered and released. Disputed orders freeze release and retain an expiry refund path; expired funded orders refund only the hirer. No private keys are accepted, no arbitrary payee or payer substitution is allowed, and the adapter grants no authority. Read /api/workorder-contract for schemas, recovery and evidence bounds. Live funded acceptance remains a manual operator step.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
wever_workorder-releaseOnly the hirer may approve a delivered order.Reads text others write2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Warn
The input schema at properties.mandate.properties.context.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.mandate.properties.context.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
- Warn
The input schema at properties.mandate.properties.terms.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.mandate.properties.terms.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
Description, as the server wrote it Only the hirer may approve a delivered order. The AP2 gateway verifies a trusted holder-bound, unexpired, unused mandate for the exact order intent. A separate payer-signed EIP-712 escrow release mandate authorizes payment only to the immutable worker. Confirmed settlement produces signed record-backed receipts for both parties. POST /api/workorder-release. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. Order value is capped at 25.00 USDC under the existing escrow-lite rail; API access fees and escrow principal are separate. Normal states are open, funded, accepted, delivered and released. Disputed orders freeze release and retain an expiry refund path; expired funded orders refund only the hirer. No private keys are accepted, no arbitrary payee or payer substitution is allowed, and the adapter grants no authority. Read /api/workorder-contract for schemas, recovery and evidence bounds. Live funded acceptance remains a manual operator step.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: false
- openWorldHint: true
Can't be undone 47Claude Code: ask
wever_budget-controls-checkAgent Governance: submit an agent budget setup: its maximum per payment, session or daily budget with its period, permitted recipients, revocation path with when it takes effect an…Can send data out1 finding
Why it is in this group
- The input addresses people (input.recipients): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.recipients): what the tool writes goes there.
Findings
- Warn
The server marks "wever_budget-controls-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.recipients): what it sends cannot be taken back.
Description, as the server wrote it Agent Governance: submit an agent budget setup: its maximum per payment, session or daily budget with its period, permitted recipients, revocation path with when it takes effect and who can use it, and spending record with when a payment appears and who can read it. The signed report gives each of five controls (maximum per payment, period budget, permitted recipients, immediate revocation and live spending record) present, weak or missing, and names budgets that cap totals and leave recipients open, revocation that is not immediate and spending records nobody can read while money is moving. Enforcing a budget on one request is wever_agent-budget-guard and an approval against a standing limit is wever_approval-policy-check. Verdict controls_missing, controls_weak, coverage_unverified or full_coverage. POST /api/budget-controls-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-capture-checkAgent Commitments: submit the messages and work records an agent wrote with the commitment records captured from them: each source with its id, its kind (a message, an email, a cha…2 findings
Why it is in this group
- The name "wever_commitment-capture-check" says "capture": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,953 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
- Warn
The server marks "wever_commitment-capture-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_commitment-capture-check" says "capture": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Commitments: submit the messages and work records an agent wrote with the commitment records captured from them: each source with its id, its kind (a message, an email, a chat, a meeting note, a ticket or a work record), who wrote it and to whom, when, its text as written, read for promise wording such as I will send or we promise and for promises made without saying so such as leave it with me or I am on it, and optionally how many promises it makes as recorded; and each commitment record with its id, the id of the source it was made in, the promiser, the promisee, the deliverable, the deadline as a date and the event it is due on. The signed report gives every source with the promise wording read from it, the promises it makes, the records that capture them and its status as captured, partly_captured, uncaptured, implied_uncaptured, no_promise or unconfirmed, every record with its status as complete, party_missing, deliverable_missing, deadline_missing or unsourced, and counts the promises, records and uncaptured promises, and names promises left uncaptured: promises said outright with no commitment record, promises made without saying so with no record, records with no promiser or promisee, and records with nothing promised; then records with no deadline or event and records with no source, with the fix for each. This reads the promises an agent itself makes: who delegated what to whom is wever_delegation-record-check, and who owns which task in a multi-agent plan is wever_coordination-plan-check. Verdict commitments_uncaptured, commitment_capture_gaps, commitment_capture_unverified or commitments_captured. POST /api/commitment-capture-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-record-checkAgent Consent: submit the consent ledger entry a grant leaves behind and the events behind it: the entry with its id, who gave the consent, who it was given to, the data, actions, …Can send data out3 findings
Why it is in this group
- The input addresses people (input.entry.recipients): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.entry.recipients): what the tool writes goes there.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 4,099 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
- Warn
The server marks "wever_consent-record-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.entry.recipients): what it sends cannot be taken back.
Description, as the server wrote it Agent Consent: submit the consent ledger entry a grant leaves behind and the events behind it: the entry with its id, who gave the consent, who it was given to, the data, actions, recipients and purposes it covers, when it was given and ends, how it was captured, the notice version shown, its status (active, narrowed, revoked or expired) and when the status was last updated; and each event with its id, when it happened, its kind (granted, narrowed, widened, renewed, revoked, expired or used), the data it concerns and who made it happen. The signed report gives the entry, every event in time order with its status as consistent, after_end, before_grant, not_by_principal or untimed, the standing the entry records beside the standing the events support with when the consent ended, and counts the events, uses, uses after the end, uses before the grant and untimed events, and names a ledger that does not hold the consent as it stands: no record of who gave it, to whom or when, a status that misses a withdrawal, an expiry or a narrowing, a use logged after the consent ended or before it was granted, data beyond what the grant events show the principal granting and a widening by someone other than the principal; then no grant event, a grant time that differs from it, a status with no event behind it, a status time older than the event that changed it, an untimed event and no purposes, capture method, notice version or status time, with the fix for each. A use after a withdrawal counts until a later grant or renewal. This reads the record one consent leaves behind: recording and evaluating wallet-scoped permission grants and revocations is wever_agent-permission-ledger, a system’s delegation records are wever_delegation-record-check and a mandate’s current state is wever_mandate-status. Verdict consent_record_missing, consent_record_incomplete, consent_record_unverifiable or consent_record_complete. POST /api/consent-record-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-revocation-checkAgent Consent: submit the grant, the planned use, the withdrawal log and the systems that must honor it: the grant with its id, data, actions and recipients; the use with its data,…Can send data out3 findings
Why it is in this group
- The input addresses people (input.grant.recipients): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.grant.recipients): what the tool writes goes there.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 4,125 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
- Warn
The server marks "wever_consent-revocation-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.grant.recipients): what it sends cannot be taken back.
Description, as the server wrote it Agent Consent: submit the grant, the planned use, the withdrawal log and the systems that must honor it: the grant with its id, data, actions and recipients; the use with its data, action, recipients and start; each withdrawal with its id, when the principal made it, whether it withdraws the whole grant or part of it, the data, actions and recipients it withdraws and how it arrived (an empty list when there are none); when the log was read; each system that holds the data or acts on the grant with its id, name, the data it holds, whether it applied the withdrawal (applied, pending, not applied or unknown) and when; and optionally how soon a withdrawal must take hold, in hours. The signed report gives every withdrawal with what it withdraws and whether it reaches the use, every system with the withdrawals that reach it and its status as honored, not_honored, pending, unknown, late or unaffected, and counts the withdrawals whole and in part, those reaching the use and the systems honored, not honored, pending, unknown and late, and names a grant revoked for the use: a whole withdrawal, a partial withdrawal of data, an action or a recipient the use relies on, and a system holding the withdrawn data that has not applied it; then a partial withdrawal the use stays outside, a system still applying it, a system whose state is not known, a system slower than the target and a log read more than a day before the use, with the fix for each. A system with no data listed is reached by every withdrawal. This reads the principal’s withdrawals and where they must hold: pulling back a delegation is wever_delegation-revocation-check, revoking or halting a trading or treasury mandate is wever_mandate-revoke or wever_mandate-halt and whether a removal request clears stored memory everywhere it lives is wever_memory-deletion-check. Verdict consent_revoked, consent_revocation_partial, consent_revocation_unchecked or consent_standing. POST /api/consent-revocation-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-scope-checkAgent Consent: submit the principal’s grant and the planned use: the grant with its id, who gave it, who it was given to, the data it covers by category, the actions it allows, the…Can send data out3 findings
Why it is in this group
- The input addresses people (input.grant.recipients): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.grant.recipients): what the tool writes goes there.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 3,936 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
- Warn
The server marks "wever_consent-scope-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.grant.recipients): what it sends cannot be taken back.
Description, as the server wrote it Agent Consent: submit the principal’s grant and the planned use: the grant with its id, who gave it, who it was given to, the data it covers by category, the actions it allows, the recipients who may receive the data or the result, the data, actions and recipients it rules out by name and its wording, read for catch-all wording such as any data, and other information or our partners; and the use with who will run it, the data it touches, the action it takes, the recipients of the data or the result and its description. The signed report gives the grant with any catch-all wording, every item the use relies on with its dimension (data, action or recipient) and its status as covered, excluded, catch_all, outside or unread, and counts the items covered, outside, excluded, reached by catch-all wording and unread, and names a use out of scope: data, an action or a recipient outside the lists the grant states, anything the grant rules out by name and a use run by another party than the one the grant was given to; then items reached only through catch-all wording and a grant that names no grantee, with the fix for each. Items compare word for word, a plural read as its singular; a list the grant does not state leaves the items in it unread. This reads one planned use against the grant, item by item: one proposed sharing of live context against the owner’s sharing rules is wever_context-sharing-check, a delegation’s written bounds are wever_delegation-scope-check, what a memory system keeps at all is wever_memory-scope-check and a wallet-scoped permission record evaluated for an exact subject, action and resource is wever_agent-permission-ledger. Verdict consent_out_of_scope, consent_scope_partial, consent_scope_unclear or consent_in_scope. POST /api/consent-scope-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-transfer-checkAgent Consent: submit the grant’s onward rule and one planned transfer: the grant with its id, what it says about passing the data on (barred, to named recipients, with notice, wit…Can send data out3 findings
Why it is in this group
- The name "wever_consent-transfer-check" says "transfer": a word for a change that another call does not undo.
- The input addresses people (input.transfer.recipient): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.transfer.recipient): what the tool writes goes there.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 4,467 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
- Warn
The server marks "wever_consent-transfer-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_consent-transfer-check" says "transfer": a word for a change that another call does not undo.
Description, as the server wrote it Agent Consent: submit the grant’s onward rule and one planned transfer: the grant with its id, what it says about passing the data on (barred, to named recipients, with notice, with fresh consent, or allowed), the recipients it names, what it says about passing the work to a sub-agent (barred, to named ones, or allowed) and its wording, read for wording that bars onward sharing such as never shared or not sold, or opens it such as our partners; and the transfer with the recipient, its kind (a third party, a sub-agent, a processor, an affiliate or the principal), the data that would pass, whether the recipient keeps to the grant’s purpose, whether it may pass the data on again or keep it after the task, whether the principal was told or agreed, and what binds the recipient to the grant. The signed report gives the rule that applies (the sub-agent rule for a sub-agent, else the onward rule, else wording that bars sharing) with where it came from, the transfer, and each condition (onward rule, recipient named, principal told, fresh consent, same purpose, no onward passing, deleted after the task, bound in writing) as met, unmet, unstated or not_required, with counts, and names a transfer the grant bars: a barred transfer, a recipient the grant does not name where it names recipients, no fresh consent where the grant requires it and a recipient free to pass the data on where the grant does not allow it; then the principal not yet told where the grant requires it, onward sharing opened only by wording, a recipient free to use the data for another purpose or keep it, nothing in writing binding the recipient and unbounded onward passing, with the fix for each. Returning data to the principal is always within the grant. This reads one transfer under the original grant: where stored memory can cross to another agent, tool or principal is wever_memory-sharing-check, what a handoff format carries is wever_context-transfer-check, the access a receiving agent gets is wever_handoff-scope-check, authority passed down a chain of delegations is wever_delegation-chain-check and who can see an agent’s traces and logs is wever_observability-access-check. Verdict consent_transfer_barred, consent_transfer_conditional, consent_transfer_unspecified or consent_transferable. POST /api/consent-transfer-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_context-transfer-checkAgent Handoffs: submit a handoff format: each field it carries with the part of the context it holds (the goal, the current state, the decisions already made, the limits the task r…1 finding
Why it is in this group
- The name "wever_context-transfer-check" says "transfer": a word for a change that another call does not undo.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_context-transfer-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_context-transfer-check" says "transfer": a word for a change that another call does not undo.
Description, as the server wrote it Agent Handoffs: submit a handoff format: each field it carries with the part of the context it holds (the goal, the current state, the decisions already made, the limits the task runs under or the open questions) and whether the format refuses a handoff without it, plus how the receiving agent fills each part the format leaves out (asks the sender, reads a shared record, reconstructs it from history or assumes it). The signed report says part by part what travels with the task and what the receiver is left to reconstruct or assume, and names every part the format leaves out that the receiver assumes or has no source for, every part it reconstructs from history or must ask the sender for after the handoff, and every part carried only in an optional field, with the fix for each. This reads the format every handoff uses, not one job’s state: one job’s state assembled into a resume pack is wever_job-handoff-pack, the records of handoffs made are wever_handoff-record-check and whether sharing context with a recipient needs consent is wever_context-sharing-check. Verdict receiver_guesses, transfer_gaps, transfer_unverified or context_travels. POST /api/context-transfer-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-escalation-checkAgent Coordination: submit how a group of agents handles a disagreement or a stall, as its escalation path, its past escalations, or both: where the path is written (a runbook, the…Can send data out1 finding
Why it is in this group
- The input addresses people (input.path.recipient): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.path.recipient): what the tool writes goes there.
Findings
- Warn
The server marks "wever_coordination-escalation-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.path.recipient): what it sends cannot be taken back.
Description, as the server wrote it Agent Coordination: submit how a group of agents handles a disagreement or a stall, as its escalation path, its past escalations, or both: where the path is written (a runbook, the work plan, configuration, only inside an agent prompt, or nowhere), who receives an escalation by name and whether that is the principal, a lead, an operator or other, whether it covers disagreements, stalls or both, what starts it (a time limit, a number of attempts, a threshold, an agent’s judgment or nothing) with its value, and whether escalations are recorded; and each past escalation with its reason, who raised it and when, who it went to, when it was resolved, how it ended (decided, withdrawn, timed out or unresolved) and the decision recorded. The signed report gives the path, every escalation and the outcomes counted, and names an escalation that is missing: a path written nowhere, a path that exists only in a prompt with no named recipient, and a written path with no recipient; then a prompt-only path, no trigger, a judgment trigger, a limit with no value, disagreements or stalls left out, escalations unrecorded, past escalations ending with no decision and escalations sent to the wrong recipient, with the fix for each. An incident escalation policy is wever_escalation-threshold-check and where a refused or silent handed-off task goes is wever_handoff-failure-check. Verdict escalation_missing, escalation_gaps, escalation_unverified or escalation_named. POST /api/coordination-escalation-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_double-charge-checkAgent Purchases: submit the calls a buyer ordered once (id, endpoint, price, when ordered) and the call sequence as recorded: every 402 challenge, settlement, refund, delivery and …1 finding
Why it is in this group
- The name "wever_double-charge-check" says "charge": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_double-charge-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_double-charge-check" says "charge": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Purchases: submit the calls a buyer ordered once (id, endpoint, price, when ordered) and the call sequence as recorded: every 402 challenge, settlement, refund, delivery and failure, each with the call it belongs to, its time, amount, settlement reference and status. The signed report reads the charge record call by call and names every call settled more than once with each settlement reference and the extra amount standing, every 402 that arrived after a call had settled asking the buyer to pay again and whether the buyer paid again, every settlement for a call never ordered, and every duplicate a refund answered. One settlement recorded twice under one reference counts once. A retry design read for where a second charge could happen is wever_retry-safety-check, a timed-out checkout against merchant orders is wever_duplicate-order-check and expected settlements against received transfers are wever_settlement-reconciliation-check. Verdict double_charged, repay_requested, charges_unverified or charged_once. POST /api/double-charge-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_energy-demand-charge-checkAgent Energy: submit a commercial bill’s demand charge with interval usage and the tariff’s demand rules: optionally the tariff with its name, demand rate per kW, interval length i…2 findings
Why it is in this group
- The name "wever_energy-demand-charge-check" says "charge": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,746 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
- Warn
The server marks "wever_energy-demand-charge-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_energy-demand-charge-check" says "charge": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Energy: submit a commercial bill’s demand charge with interval usage and the tariff’s demand rules: optionally the tariff with its name, demand rate per kW, interval length in minutes (15 when left out), the peak window with its days (weekdays or every day) and its opening and closing times, the ratchet with its share of the highest earlier peak and its look-back months, and the power factor threshold; the intervals with each start in local meter time and its kW or kWh (an empty list when none); earlier months’ peaks for a ratchet; the measured power factor; the bill with its period, billed kW, printed demand rate, demand charge, and whether it says a ratchet or a power factor adjustment set the billed demand; and optionally the tolerance in kW (0.1 when left out). The signed report gives the intervals with the expected and supplied counts, the window and the peak inside it and at any time, the demand with the period’s own peak, the ratchet floor, the power factor, the recomputed and billed kW and their difference, the charge with the tariff and billed rates and the recomputed and billed charge, and counts them, and names an overstated charge: billed demand above the highest counted interval, a peak taken outside the window, a ratchet or power factor adjustment the rules do not call for, a rate above the tariff and a charge above the recomputed charge; then billed demand, a rate or a charge below the recomputed figures and missing intervals, with the fix for each and what would change the verdict. No intervals, rate, look-back history, power factor or billing period leave the charge unverified. This recomputes one demand charge: the bill’s other lines against the tariff and usage are wever_energy-bill-check, an invoice against its job is wever_invoice-check and fees charged against a stated schedule are wever_fee-reconciliation-check. Verdict energy_demand_overstated, energy_demand_partial, energy_demand_unverified or energy_demand_correct. POST /api/energy-demand-charge-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-request-checkAgent Feedback: submit one feedback request on a specific piece of work, as recorded: optionally the request itself with the work it names, its text as sent, when it went out and t…Can send data out2 findings
Why it is in this group
- The name "wever_feedback-request-check" says "request": a word for a change.
- The input addresses people (input.recipients): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.recipients): what the tool writes goes there.
Findings
- Info
The description is 2,274 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
- Warn
The server marks "wever_feedback-request-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_feedback-request-check" says "request": a word for a change.
Description, as the server wrote it Agent Feedback: submit one feedback request on a specific piece of work, as recorded: optionally the request itself with the work it names, its text as sent, when it went out and the time to respond by; the criteria it asks givers to judge the work against; and each party it went to, with whether that party saw or used the work and their relation to it (a user, a reviewer, the principal, a peer, a stakeholder or other). The signed report gives the request with any open-invitation wording it uses and the window to respond in days, every recipient as saw_work, never_saw_work or unstated, and the criteria, and names a request that leaves the loop unfocused: no specific work named, or work named only as everything or my work; no criteria, or an open invitation such as any feedback welcome in their place; and recipients who never saw or used the work; then a request with no time to respond by and a request with no named recipients, with the fix for each. This reads one request for feedback before any feedback arrives, not a formal evaluation: an evaluation’s design is wever_evaluation-design-check and scoring an agent against a rubric is wever_evaluation-scoring-check, and how a group of agents plans and splits shared work is wever_coordination-plan-check. Verdict feedback_request_unfocused, feedback_request_gaps, feedback_request_unverified or feedback_request_focused. POST /api/feedback-request-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-limits-checkAgent Insurance: submit one policy’s limits with the stated exposure: the policy with its currency, the per-occurrence and aggregate limits as recorded and each sublimit with its c…2 findings
Why it is in this group
- The input carries an amount (input.exposure.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,811 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
- Warn
The server marks "wever_insurance-limits-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.exposure.amount) and a currency, so it moves money.
Description, as the server wrote it Agent Insurance: submit one policy’s limits with the stated exposure: the policy with its currency, the per-occurrence and aggregate limits as recorded and each sublimit with its category; the exposure with the category of loss it falls in, the stated exposure for one occurrence, its currency and description; each prior claim in the policy period with its id, what it paid and what is reserved against the aggregate, its date and category, as recorded; and optionally a contract with the contracting party, its currency and the per-occurrence, aggregate and category minimums it requires. The signed report gives the policy, the exposure, the limit one occurrence draws on as the lower of the per-occurrence limit and the sublimit on the exposure’s category with which one binds, the shortfall and the exposure’s share of the limit, the aggregate with the prior claims, the amount eroded, the amount left and the share used, and every contract minimum against the figure the policy holds, and counts them, and names short limits: a per-occurrence limit below the stated exposure, a sublimit that caps the exposure below it, an aggregate eroded below the exposure by prior claims, and limits under a contract’s required minimums; then an exposure at 90% or more of its limit, an aggregate half used or more and an exposure or contract stated in another currency, with the fix for each and what would change the verdict. No per-occurrence limit, no stated exposure, no exposure category where the policy has sublimits, no prior claims record or no aggregate where erosion or a contract needs one leaves the limits unverified. This reads one policy’s limits against the stated exposure: a counterparty’s certificate limits against required minimums read as diligence evidence are wever_insurance-evidence-check, an agent’s own spend caps are wever_exposure-cap-check and one planned action’s worst case against its limits is wever_risk-exposure-check. Verdict insurance_limits_short, insurance_limits_partial, insurance_limits_unverified or insurance_limits_sufficient. POST /api/insurance-limits-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_invoice-payment-matchCommerce Assurance: submit one invoice, one payment or transfer record and an optional receipt.Can send data out1 finding
Why it is in this group
- The name "wever_invoice-payment-match" says "payment": a word that usually means money moves, which cannot be taken back.
- The input carries an amount (input.invoice.amount) and a currency, so it moves money.
- The input addresses people (input.payment.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.payment.to): what the tool writes goes there.
Findings
- Warn
The server marks "wever_invoice-payment-match" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_invoice-payment-match" says "payment": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Commerce Assurance: submit one invoice, one payment or transfer record and an optional receipt. The signed report compares amount, currency, payer, payee, reference and date, itemizing every difference: a short or over payment with the amount, a currency that differs, a payer or payee that differs by name or account, a missing or different reference, and a payment before issue or after the due date. Verdict exact, partial or mismatch. POST /api/invoice-payment-match. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-capture-checkAgent Learning: submit the experience episodes an agent went through with the lesson records captured from them: each episode with its id, its kind (a task, a success, a failure, a…2 findings
Why it is in this group
- The name "wever_learning-capture-check" says "capture": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,105 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
- Warn
The server marks "wever_learning-capture-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_learning-capture-check" says "capture": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Learning: submit the experience episodes an agent went through with the lesson records captured from them: each episode with its id, its kind (a task, a success, a failure, an incident, a correction, a surprise or a near miss), when it happened, the context it happened in, its log as written, read for lesson wording such as next time, lesson learned, should have or root cause, and optionally how many lessons it carries as recorded; each lesson record with its id, the id of the source episode it records, the date of the experience, when the record was written, the context it was learned in and the lesson as recorded; and optionally the days a record may follow its experience (7 when left out). The signed report gives every episode with the lesson wording read from it, the lessons it carries, the records that capture them and its status as captured, partly_captured, uncaptured, no_lesson or unconfirmed, every record with its days after the experience and its status as complete, empty, unsourced, undated, no_context or late, and counts the lessons carried, the records and the lessons left uncaptured, and names experience left uncaptured: episodes that carry a lesson with no lesson record and records that state no lesson; then records with no source episode, no date or no context, and records written long after the experience, with the fix for each. This reads the agent’s own experience captured as lessons: whether outside feedback on one piece of work was incorporated is wever_feedback-incorporation-check, promises captured from messages are wever_commitment-capture-check, and incident claims read against a postmortem are wever_incident-evidence-check. Verdict experience_not_captured, capture_incomplete, capture_informal or experience_captured. POST /api/learning-capture-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-transfer-checkAgent Learning: submit an agent’s lessons with the new tasks and contexts they were considered for: each lesson with its id, the lesson as stated, read for wording such as always, …2 findings
Why it is in this group
- The name "wever_learning-transfer-check" says "transfer": a word for a change that another call does not undo.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,855 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
- Warn
The server marks "wever_learning-transfer-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_learning-transfer-check" says "transfer": a word for a change that another call does not undo.
Description, as the server wrote it Agent Learning: submit an agent’s lessons with the new tasks and contexts they were considered for: each lesson with its id, the lesson as stated, read for wording such as always, never or in every case, the context it was learned in, where it holds and the cases it leaves out, as stated; and each new task with its id, when it ran, the task as recorded, its context, the id of the lesson considered for it, whether it falls inside the lesson’s stated scope and whether the lesson was carried into it. The signed report gives every task with its lesson and its status as carried, held_back, dropped, overextended or unconfirmed, every lesson with its scope, its limits, any always or never wording and its status as scoped, overextended, dropped, unscoped, absolute or unconfirmed, and counts the lessons with a stated scope and the tasks carried, held back, dropped and overextended, and names lessons carried past their scope: tasks outside a lesson’s stated scope that the lesson was carried into; then tasks inside a lesson’s scope that went without it, lessons considered for new work with no stated scope, and lessons worded to hold everywhere with no stated limits, with the fix for each. This reads the agent’s own lessons carried into new work within their scope: what a handoff format carries to a receiving agent is wever_context-transfer-check, and stored memory crossing to another agent or principal is wever_memory-sharing-check. Verdict transfer_overreaching, transfer_partial, no_transfer or transfer_scoped. POST /api/learning-transfer-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-return-checkAgent Migration: submit whether one move can be reversed, as its return path, the steps that cannot be undone, or both: how the plan presents the move (reversible or one way), whet…1 finding
Why it is in this group
- The description says "cannot be undone".
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_migration-return-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "cannot be undone".
Description, as the server wrote it Agent Migration: submit whether one move can be reversed, as its return path, the steps that cannot be undone, or both: how the plan presents the move (reversible or one way), whether a way back to the source is written down, how the agent moves back, how long the way back stays open after cutover, what returning costs and whether the return has been rehearsed; and each step that cannot be undone once it happens (source data deleted, a name or handle released, funds spent, a key revoked, the source account closed, a contract ended or other) with how many days after cutover it happens and whether the plan marks it as permanent. The signed report gives the return path, its window in days with the steps that close it early, and every permanent step with its status, and names a one-way door presented as reversible: a move presented as reversible with no written way back, and a permanent step such as deleting the source data that happens before the return window closes; then an honest one-way move, a way back with no deadline, no stated cost or no rehearsal, and permanent steps the plan does not mark, with the fix for each. The general rollback path for one action’s external effects is wever_rollback-path-check, and halting a running agent system is wever_kill-switch-check. Verdict return_path_closed, return_path_gaps, return_path_unverified or return_path_open. POST /api/migration-return-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_money-fit-checkAgent Readiness: submit a task an agent must pay for (amount, counterparty type, whether it requires a refund path, whether a party other than the payer must carry loss) and the mo…1 finding
Why it is in this group
- The description says "irreversible".
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_money-fit-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "irreversible".
Description, as the server wrote it Agent Readiness: submit a task an agent must pay for (amount, counterparty type, whether it requires a refund path, whether a party other than the payer must carry loss) and the money options available with their stated properties (kind such as stablecoin, tokenized deposit or agent card, reversibility, refund path, redemption, who carries loss, limits, accepted by, fee). The signed report names the option that fits and why, each option ruled out with the property that rules it out (irreversible where a refund path is required, redemption risk, liability mismatch, limits, acceptance), and every property the material leaves unstated. Payment rails are wever_payment-rail-check. Verdict no_option_fits, fit_unconfirmed or fit_found. POST /api/money-fit-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-mandate-checkAgent Negotiation: submit one negotiation mandate: the agent and its principal, who authorized it, when and how that is recorded (signed, written, a message, spoken or not at all),…Can send data out1 finding
Why it is in this group
- The input addresses people (input.escalation.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.escalation.to): what the tool writes goes there.
Findings
- Warn
The server marks "wever_negotiation-mandate-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.escalation.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Negotiation: submit one negotiation mandate: the agent and its principal, who authorized it, when and how that is recorded (signed, written, a message, spoken or not at all), the deal it covers and its expiry, each walk-away limit with its term, kind, direction (at most, at least or exactly), value and whether reaching it ends the negotiation, the fixed terms, the flexible terms with how far each may move, who decides beyond the mandate, and optionally the moment to read it at. The signed report says limit by limit and term by term whether the agent is bounded and whether the mandate is in force, and names no authorizer, no limits, no price limit, limits with no value, in open wording (such as any or no cap) or with no direction, unbounded flexible terms, an expired mandate and use before authorization, then limits no value can meet together, terms both fixed and flexible or never classified, no walk-away point, no expiry, an unrecorded authorization, no deal named and nobody to escalate to, with the fix for each. One delegation’s scope is wever_delegation-scope-check, access against an authority grant is wever_authority-scope-check and issuing a signed mandate is wever_delegated-authority. Verdict mandate_unbounded, mandate_gaps, mandate_unverified or mandate_bounded. POST /api/negotiation-mandate-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-settlement-checkAgent Negotiation: submit one agreed deal and its settlement: the parties, the agreed price, currency, method (stablecoin, card, bank transfer, escrow, invoice, cash or other), net…1 finding
Why it is in this group
- The input carries an amount (input.settlement.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_negotiation-settlement-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.settlement.amount) and a currency, so it moves money.
Description, as the server wrote it Agent Negotiation: submit one agreed deal and its settlement: the parties, the agreed price, currency, method (stablecoin, card, bank transfer, escrow, invoice, cash or other), network, payment and delivery due and the order of delivery and payment (payment first, delivery first, both at once, escrow or milestones), the settlement’s amount, currency, method, network, timing, order and escrow holder, and each settlement step with the party that does it. The signed report reads the settlement against the agreement field by field and step by step, and names a different amount, currency, method, network or order, payment later than due, no stated order and every step no party owns, then no network for a stablecoin payment, escrow with no holder, no due dates and steps after a step the list lacks, with the fix for each. Expected settlements against transfers are wever_settlement-reconciliation-check, a payment instruction is wever_payment-instruction-check and a settlement claim against chain evidence is wever_settlement-proof-check. Verdict cannot_settle, settlement_gaps, settlement_unverified or settles_as_written. POST /api/negotiation-settlement-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_observability-metric-checkAgent Observability: submit the metrics an operator reports, each with its name, its kind (a success rate, a failure rate, an average or 95th percentile latency, an average cost pe…Can send data out1 finding
Why it is in this group
- The input addresses people (input.period.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.period.to): what the tool writes goes there.
Findings
- Warn
The server marks "wever_observability-metric-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.period.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Observability: submit the metrics an operator reports, each with its name, its kind (a success rate, a failure rate, an average or 95th percentile latency, an average cost per task or a task count), its figure, its unit, the number of runs it says it counts, its window and the run statuses it says it leaves out, together with the run records behind them, each with its id, date, status (succeeded, failed, partial, timed out or cancelled), latency and cost, and optionally the reporting period. The signed report recomputes every figure from the records over its stated window with only its named exclusions, and when it differs tries every set of unnamed failing statuses, smallest first, so it names numbers that cannot be derived: a figure the records do not give, a success rate with no stated denominator, averages and rates that hold only with the failures left out, and metrics computed over a window that excludes the days the runs went worse; then a stated run count the records do not give, a shorter window that hides nothing worse, no window and undated records, with the fix for each. An evaluation’s claims against its measured results are wever_evaluation-result-check, headline payment numbers with no commerce behind them are wever_volume-quality-check and a reputation claim against its job record is wever_reputation-claim-check. Verdict metrics_underivable, metric_gaps, metrics_unverified or metrics_derivable. POST /api/observability-metric-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_observability-report-checkAgent Observability: submit one observability summary written for a principal: its title and audience, the headline as written with the uptime and success rate it reports and the p…Can send data out1 finding
Why it is in this group
- The input addresses people (input.summary.period.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.record.to): what the tool writes goes there.
Findings
- Warn
The server marks "wever_observability-report-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.summary.period.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Observability: submit one observability summary written for a principal: its title and audience, the headline as written with the uptime and success rate it reports and the period it covers; whether it states the failed runs, the gaps in trace and log coverage and the downtime, where each sits (beside the headline, in a later section, in an appendix or in a separate document) and how many or how many minutes it reports; and optionally the underlying record’s runs, failures, coverage gaps and downtime minutes with its period. The signed report says what sits beside the headline and checks it against the record, recomputing the uptime from the downtime over the period and the success rate from the runs and failures, and names a summary that leaves out the failures, the gaps in coverage or the downtime, or reports none against a record that holds some; then amounts below the record’s, disclosures in an appendix or a separate document, an uptime or success rate the record does not give, a period that differs from the record’s and no period, with the fix for each. An evaluation summary’s disclosures are wever_evaluation-report-check, a reputation summary for a buyer is wever_reputation-report-check and whether a failure notice reaches the right party is wever_failure-notice-check. Verdict headline_only, report_gaps, report_unverified or report_complete. POST /api/observability-report-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-limits-checkAgent Onboarding: submit the spending limits a new agent runs under, before its first task: optionally the agent with its id and when it runs its first task; the job with whether i…Can send data out3 findings
Why it is in this group
- The input addresses people (input.handoff.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.handoff.to): what the tool writes goes there.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
- Info
The description is 4,623 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
- Warn
The server marks "wever_onboarding-limits-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.handoff.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Onboarding: submit the spending limits a new agent runs under, before its first task: optionally the agent with its id and when it runs its first task; the job with whether it spends or moves money, the largest single amount it legitimately needs, its unit, the merchants it pays and the kinds of task it runs; each limit with its id, name, kind (a cap on one action, one task, one merchant, over a period or in total), amount, unit, the task or merchant it applies to, where it is enforced (code, a gateway, a wallet policy or the platform outside the agent, only the prompt, or nowhere) and when it was set (an empty list when none); and the human handoff with the amount above which a person decides, its unit, the named person and where it is enforced. The signed report gives every limit with its status as set, in_prompt, set_late, blocks_job, far_above_job, other_unit or unread, the handoff, the coverage with the tasks and merchants the job spans and whether a limit is scoped per task and per merchant, and counts the limits, those in code, those only in the prompt, those set late and the caps on one action, and names missing limits: a job that moves money with no limit, a limit only in the prompt or enforced nowhere and a limit set after the first task; then no cap on one action or task, a cap below the job’s largest legitimate amount or ten times above it, several tasks or merchants with no limit scoped to one, no handoff line, no named person above it, a line above the cap so amounts between are refused and never reach a person, and a handoff only in the prompt, with the fix for each. This reads that limits were put in place before the first task, the limits the wever_limit checks review while work runs: where each limit is enforced is wever_limit-substrate-check, which limit governs each kind of job is wever_limit-scope-check, what hitting a limit does is wever_breach-path-check, a budget’s five controls are wever_budget-controls-check, a spending policy’s stated caps are wever_spending-policy-check, one request against a budget is wever_agent-budget-guard, a cap’s scope, window and breach rule is wever_exposure-cap-check and one planned action’s worst case against the limits is wever_risk-exposure-check. Verdict onboarding_limits_missing, onboarding_limits_partial, onboarding_limits_unverified or onboarding_limits_set. POST /api/onboarding-limits-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-readiness-checkAgent Onboarding: submit whether a new agent is ready for its first task: the job with its id, the written job, read for wording that leaves it open-ended such as help with anythin…Can send data out3 findings
Why it is in this group
- The input addresses people (input.escalation.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.escalation.to): what the tool writes goes there.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
- Info
The description is 4,493 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
- Warn
The server marks "wever_onboarding-readiness-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.escalation.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Onboarding: submit whether a new agent is ready for its first task: the job with its id, the written job, read for wording that leaves it open-ended such as help with anything or other duties as assigned, the tasks it covers, the named person who owns the agent’s work, the first task, how the owner tells it is done, read for wording that leaves done to opinion such as looks good, how the first task is supervised (every action reviewed, a sample reviewed, exceptions only, or not at all) and when it runs; and the escalation path with the named person the agent escalates to, how it reaches them, what counts as outside the job and how soon that person responds. The signed report gives the job with whether the first task is among the tasks the job covers and any open-ended or vague wording, the escalation path, every requirement (a written job, a named owner, a first task the job covers, a test for when it is done, a supervision level, an escalation path) as ready, gap, missing or unknown, and counts them, and names an agent not ready: no written job, owner or first task, an owner named as a team, a placeholder or the agent itself, a first task outside the written job and an escalation path with no one on it; then open-ended job wording, no done test or one that leaves done to opinion, no supervision, escalation to a placeholder and no stated trigger or route, with the fix for each. No task list, supervision level or escalation path leaves readiness unchecked. This reads readiness to start the first task, with the job the wever_goal checks read as the record the first task is read against: the outcome a goal records is wever_goal-statement-check and its working scope wever_goal-scope-check, the rule set an operator hands an agent is wever_adoption-readiness-check, a production agent’s named owner and fallback are wever_incident-ownership-check, the accountable person for each outcome is wever_ownership-map-brief, an escalation policy’s triggers are wever_escalation-threshold-check and the consequences of one planned action are the wever_risk checks such as wever_risk-exposure-check. Verdict onboarding_not_ready, onboarding_readiness_partial, onboarding_readiness_unchecked or onboarding_ready. POST /api/onboarding-readiness-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_payment-authority-inspectorPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- The name "wever_payment-authority-inspector" says "payment": a word that usually means money moves, which cannot be taken back.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...t-authority-inspector. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Inspect a canonical Base USDC transaction receipt for the configured Wever recipient and return the observed payment details. Operating boundary: Supply the receipt reference. Read the chain, recipient, amount and verification result. An agent links the receipt to its work record; a human inspects the same transaction details. POST /api/payment-authority-inspector. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current money service follows its published inspection, policy evaluation or private invoice contract. Payment is an API access fee and grants no spending authority.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_payment-instruction-checkCommerce Assurance: submit a payment instruction as fields (network, asset, recipient, amount, currency, reference, memo, expires_at, payee), as text or both, with an optional invo…Can send data out1 finding
Why it is in this group
- The name "wever_payment-instruction-check" says "payment": a word that usually means money moves, which cannot be taken back.
- The input carries an amount (input.instruction.amount) and a currency, so it moves money.
- The input addresses people (input.instruction.recipient): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.instruction.recipient): what the tool writes goes there.
Findings
- Warn
The server marks "wever_payment-instruction-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-instruction-check" says "payment": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Commerce Assurance: submit a payment instruction as fields (network, asset, recipient, amount, currency, reference, memo, expires_at, payee), as text or both, with an optional invoice or order and optional payer policy (allowed networks and assets, maximum amount, recipient allowlist, reference and expiry rules). The signed report restates the instruction and checks it for internal consistency and against the invoice and policy: the network, the asset and its token address, the format of the recipient address on that network (with the EIP-55 checksum for mixed-case EVM addresses), the amount and its precision, the currency, the reference or memo and the expiry. Verdict consistent, mismatches_found or elements_missing. POST /api/payment-instruction-check. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_payment-policy-checkAgent preflight: check one x402 payment challenge against your spending policy before you pay.1 finding
Why it is in this group
- The name "wever_payment-policy-check" says "payment": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_payment-policy-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-policy-check" says "payment": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent preflight: check one x402 payment challenge against your spending policy before you pay. Input challenge (the 402 response body, one payment requirements object or the base64 PAYMENT-REQUIRED header, version 1 or 2) and policy (per_call_cap and daily_cap in whole units of the asset, optional spent_today, allowed_networks, allowed_assets, optional recipient_allowlist and asset_decimals). The signed report gives a pay or refuse verdict, the reason for every failed rule (challenge schema, network, asset, recipient, per-call cap, daily cap and a stated authorization window), and the amount, asset, network, recipient and expiry restated from the challenge itself. It evaluates the challenge only and never pays. POST /api/payment-policy-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_payment-rail-checkAgent Treasury: submit two or more payment rail options (name, networks or currencies, fee shape, minimum and maximum amounts, settlement time and any stated limits) and one paymen…1 finding
Why it is in this group
- The name "wever_payment-rail-check" says "payment": a word that usually means money moves, which cannot be taken back.
- The input carries an amount (input.requirement.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_payment-rail-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-rail-check" says "payment": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Treasury: submit two or more payment rail options (name, networks or currencies, fee shape, minimum and maximum amounts, settlement time and any stated limits) and one payment requirement (amount, currency, destination type, network, deadline). The signed report returns fits, fits_with_caveats or does_not_fit for each rail with the deciding facts itemized, computes each fee from its stated shape and names the best-fitting rail. Supplied options only. Verdict rail_fits, fits_with_caveats or no_rail_fits. POST /api/payment-rail-check. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_payment-recovery-checkJob Records: submit one payment proof or settled-payment record (amount, service or endpoint, transaction reference, timestamp) with your account of the lost or missing result and …Can send data out1 finding
Why it is in this group
- The name "wever_payment-recovery-check" says "payment": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a URL (input.payment.endpoint): a request to a URL can carry data to whoever runs it.
Findings
- Warn
The server marks "wever_payment-recovery-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-recovery-check" says "payment": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Job Records: submit one payment proof or settled-payment record (amount, service or endpoint, transaction reference, timestamp) with your account of the lost or missing result and any service response records you hold. The signed report ties each response to the payment by its reference or by endpoint and time, says whether the records show the result delivered, not delivered or unresolved, and names the safe next step they support: re_request_same_reference, present_proof_to_seller or treat_as_unresolved. The report never pays again and the seller is never contacted. Verdict delivered, not_delivered or unresolved. POST /api/payment-recovery-check. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_payout-change-checkAgent Operations: submit one payout or fee configuration as originally agreed (destination, network, asset, fee rate, fixed fee, caps and scope) with its approval record and the ag…1 finding
Why it is in this group
- The name "wever_payout-change-check" says "payout": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_payout-change-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payout-change-check" says "payout": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Operations: submit one payout or fee configuration as originally agreed (destination, network, asset, fee rate, fixed fee, caps and scope) with its approval record and the agreement’s change rules, a proposed change and any consent record for it. The signed report says whether the change can take effect under the original agreement, needs the counterparty’s fresh approval first (naming what changed: new destination or payout route, higher fee, wider scope) or is outside the agreement entirely, change by change. Configuration-change governance: payout readiness for a single payment is wever_payout-readiness-check. Verdict outside_agreement, counterparty_approval_required or can_take_effect. POST /api/payout-change-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_payout-readiness-checkCommerce Assurance: submit a payout request (recipient, wallet, network, amount, currency, purpose, requester) with its supporting invoice, documents, approvals and an optional pol…Can send data out1 finding
Why it is in this group
- The name "wever_payout-readiness-check" says "payout": a word that usually means money moves, which cannot be taken back.
- The input carries an amount (input.request.amount) and a currency, so it moves money.
- The input addresses people (input.request.recipient): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.request.recipient): what the tool writes goes there.
Findings
- Warn
The server marks "wever_payout-readiness-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payout-readiness-check" says "payout": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Commerce Assurance: submit a payout request (recipient, wallet, network, amount, currency, purpose, requester) with its supporting invoice, documents, approvals and an optional policy (required approvals, approver roles, required documents, maximum amount). The signed report checks the recipient, the wallet format, the amount against the invoice and policy, the purpose, the supporting documents and approval coverage, and returns ready or not_ready with every missing or inconsistent element named. The report is the readiness record for whoever runs the payout. POST /api/payout-readiness-check. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_policy-fit-checkDiligenceOps: submit a proposed agent action in plain language (optionally with its amount) and a policy or mandate text.1 finding
Why it is in this group
- The input carries an amount (input.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_policy-fit-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.amount) and a currency, so it moves money.
Description, as the server wrote it DiligenceOps: submit a proposed agent action in plain language (optionally with its amount) and a policy or mandate text. The signed report says whether the action fits: blocked, needs_condition, needs_detail, fits or not_addressed, citing each clause that allows, limits, requires or blocks it, comparing amount caps and approval thresholds exactly, and naming what the policy is silent on. POST /api/policy-fit-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_procurement-decision-briefCommerce Assurance: submit two to six vendor proposals as fields (price, billing, setup fee, term, service levels, security, data terms, liability, renewal, support) or text, with …1 finding
Why it is in this group
- The input carries an amount (input.budget.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_procurement-decision-brief" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.budget.amount) and a currency, so it moves money.
Description, as the server wrote it Commerce Assurance: submit two to six vendor proposals as fields (price, billing, setup fee, term, service levels, security, data terms, liability, renewal, support) or text, with optional priorities and budget. The signed brief normalizes the comparison across price (annualized, first year and over the term), term, service levels, security and data terms, liability, renewal and support, names each vendor’s evidence gaps, and closes with a recommendation, its rationale and the conditions that would change it. Verdict recommended or not_comparable. POST /api/procurement-decision-brief. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_purchase-record-checkAgent Purchases: submit the buyer’s own record of paid calls, each with what was ordered, the provider and endpoint, the quoted amount, the amount paid, the asset, network and dest…1 finding
Why it is in this group
- The name "wever_purchase-record-check" says "purchase": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_purchase-record-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_purchase-record-check" says "purchase": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Purchases: submit the buyer’s own record of paid calls, each with what was ordered, the provider and endpoint, the quoted amount, the amount paid, the asset, network and destination paid, the settlement reference, the receipt held, what was kept of the result (stored, a summary, only a status or nothing), the result status and the time. The signed report totals what the ledger shows was paid by currency, names every field missing from every record, with missing amounts, destinations, results and receipts first, and says for each invisible failure (price drift between quote and charge, a call charged twice, a success status carrying the wrong result, money owed back with no proof to claim it) how many records can detect it and which cannot. Approval records are wever_approval-record-check, one workflow run’s records are wever_run-record-check and one job’s payments matched to receipts are wever_job-receipt-pack, and the boutique Agent Purchase API’s own intent and order state is wever_purchase-status. Verdict failures_undetectable, record_gaps or reconstructable. POST /api/purchase-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_refund-eligibility-checkDiligenceOps: submit a purchase record (amount, currency, purchase and delivery dates, item, condition, final sale) and the merchant’s refund policy text, with an optional request …1 finding
Why it is in this group
- The name "wever_refund-eligibility-check" says "refund": a word that usually means money moves, which cannot be taken back.
- The input carries an amount (input.purchase.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_refund-eligibility-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_refund-eligibility-check" says "refund": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it DiligenceOps: submit a purchase record (amount, currency, purchase and delivery dates, item, condition, final sale) and the merchant’s refund policy text, with an optional request date and the evidence on hand. The signed report says whether a refund claim fits the policy as written, shows the deadline arithmetic step by step, and lists item conditions, exclusions, fees, the remedy and the evidence the policy expects attached. POST /api/refund-eligibility-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_refund-path-checkAgent Purchases: submit a provider’s stated refund route for failed paid calls (endpoint, process, automatic, contact-us or none, with its window, where the money goes back to and …1 finding
Why it is in this group
- The name "wever_refund-path-check" says "refund": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_refund-path-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_refund-path-check" says "refund": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Purchases: submit a provider’s stated refund route for failed paid calls (endpoint, process, automatic, contact-us or none, with its window, where the money goes back to and the proof it requires), each step of the route (endpoint, form, email, ticket, chat or dispute, where it happens, what a test of it showed and what the buyer needs to take it), the failed call (reference, amount, when it failed and when the refund is asked for) and the proof the buyer holds. The signed report gives each step as reachable, unreachable, unpublished or untested, and names every block (no route, contact-us alone, a step written down with nowhere to take it, a step tested and failing, a window already closed) and every gap (a step needing an account an agent may not hold, no stated window, credit in place of money, proof the buyer does not hold). What the seller’s refund rules say is wever_refund-rule-check and whether one purchase qualifies under a refund policy is wever_refund-eligibility-check. Verdict no_working_route, route_gaps, route_unverified or route_reachable. POST /api/refund-path-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_refund-rule-checkAgent Accountability: submit a checkout or payment setup’s refund terms (when the rule is fixed, whether the payer sees it before paying, who may request, how a request is decided …1 finding
Why it is in this group
- The name "wever_refund-rule-check" says "refund": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_refund-rule-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_refund-rule-check" says "refund": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Accountability: submit a checkout or payment setup’s refund terms (when the rule is fixed, whether the payer sees it before paying, who may request, how a request is decided and by whom, the window, the destination, the amount and how soon it is paid) and its checkout steps with what each records. The signed report says whether a refund rule exists before checkout at all, whether the refund destination is captured at checkout, and every gap that would turn a refund into a one-off negotiation, with the fix for each. One purchase against a refund policy is wever_refund-eligibility-check. Verdict no_standing_rule, negotiation_gaps or rule_before_checkout. POST /api/refund-rule-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-claim-checkAgent Reputation: submit one reputation claim made to a buyer or principal, such as a success rate, a job count, completed jobs, an average rating, a dispute rate or an on-time rat…1 finding
Why it is in this group
- The input addresses people (input.claim.window.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_reputation-claim-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.claim.window.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Reputation: submit one reputation claim made to a buyer or principal, such as a success rate, a job count, completed jobs, an average rating, a dispute rate or an on-time rate, with its claimed figure, the time window it states, the job categories it says it leaves out (failed, partial, cancelled, refunded or disputed), the number of jobs it says it counts, who it was made to and when, together with the underlying record entry by entry: each entry’s date, outcome, rating, whether it was on time and whether it was disputed. The signed report recomputes the figure from the record over the stated window with only the named exclusions, and when it differs tries every set of unnamed job categories and then the whole history, smallest set first, so it names a figure the record does not give, jobs the claim quietly excludes with the categories and counts that reproduce it, a figure that holds only outside the stated window and a stated job count that differs, then no window, a window the record does not reach back to, undated entries and a rate or average with no job count behind it, with the fix for each. A claim against supplied evidence text is wever_claim-evidence-check, a benchmark claim against its cited source is wever_benchmark-evidence-check and an evaluation’s claims against its measured results are wever_evaluation-result-check. Verdict claim_unsupported, claim_gaps, claim_unverified or claim_supported. POST /api/reputation-claim-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-report-checkAgent Reputation: submit one reputation summary written for a buyer or principal: its title and audience and the headline score as written, with whether its scale and the number of…Can send data out1 finding
Why it is in this group
- The input addresses people (input.window.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.window.to): what the tool writes goes there.
Findings
- Warn
The server marks "wever_reputation-report-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.window.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Reputation: submit one reputation summary written for a buyer or principal: its title and audience and the headline score as written, with whether its scale and the number of jobs behind it sit beside it; whether it states the disputes and the failures, where each sits (beside the headline, in a later section, in an appendix or in a separate document) and how many it reports; whether it states the record’s time window, where it sits and its dates; and optionally the underlying record’s job, dispute and failure counts with its first and last job. The signed report says what sits beside the headline score and checks it against the record, and names a summary that leaves out the disputes, the failures or the record’s time window, or reports none against a record that holds some, then counts below the record’s, disclosures in an appendix or a separate document, a window that differs from the record’s and a headline with no scale or job count, with the fix for each. An evaluation summary’s disclosures are wever_evaluation-report-check, a recommendation’s commercial interest is wever_recommendation-disclosure-check and a trust score computed from verified work is wever_agent-trust-scorecard. Verdict headline_only, report_gaps, report_unverified or report_complete. POST /api/reputation-report-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-reversibility-checkAgent Risk: submit one planned action and the effects it leaves outside itself: the action with its id, what it puts at stake, its unit and its description, read for wording that m…2 findings
Why it is in this group
- The description says "cannot be undone".
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,157 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
- Warn
The server marks "wever_risk-reversibility-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "cannot be undone".
Description, as the server wrote it Agent Risk: submit one planned action and the effects it leaves outside itself: the action with its id, what it puts at stake, its unit and its description, read for wording that marks it final such as non-refundable, all sales final or cannot be undone; each effect with its id, what it leaves behind (such as the sent order or the moved funds), how it is undone (automatically, on request, by a compensating action such as a refund, not at all, or not known), what undoing it costs, how long undoing stays possible, whether undoing it needs the counterparty to agree and whether the undo has been run and worked; how long a mistake in the action takes to notice; and optionally the share of the action’s value at which an undo counts as costly (10 when left out). The signed report gives the action with any final wording, every effect with its undo, its cost and that cost’s share of the action’s value, its window in hours and its status as reversible, irreversible, closes_early, costly, needs_counterparty, compensating, unverified or unknown, and counts the effects reversible, irreversible, closing before a mistake is noticed, costly and unverified, and names an irreversible action: an effect that cannot be undone, an undo window that closes before a mistake would be noticed and wording that marks the action final; then an undo that costs a large share of the action’s value, an undo that needs the counterparty to agree and an effect undone only by a compensating action, with the fix for each. An undo not stated, untimed, uncosted or never run leaves the action unverified. This reads one action’s undo before it is taken, the undo side the Agent Recovery checks plan for after a failure: a task system’s undo design for the effects it leaves is wever_rollback-path-check, the record a recovery leaves is wever_recovery-record-check, recovering a lost result behind a settled payment is wever_payment-recovery-check, whether one move to a new system can be reversed is wever_migration-return-check and the reversibility of a money option is wever_money-fit-check. Verdict risk_irreversible, risk_reversal_costly, risk_reversal_unverified or risk_reversible. POST /api/risk-reversibility-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-handoff-checkAgent Roles: submit a role changing holders: the handoff with its id, the role, the prior holder, the incoming holder and when it changed hands; the description with its version be…Can send data out2 findings
Why it is in this group
- The input addresses people (input.handoff.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.handoff.to): what the tool writes goes there.
Findings
- Info
The description is 3,421 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
- Warn
The server marks "wever_role-handoff-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.handoff.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Roles: submit a role changing holders: the handoff with its id, the role, the prior holder, the incoming holder and when it changed hands; the description with its version before and after, when it was rewritten and by whom, what the rewrite changed and the handoff note as written, read for wording such as as is, unchanged or same as before; the prior holder’s permissions for the role, each with whether it was revoked, and when they were revoked; the incoming holder’s acceptance with whether it was given, when and where it is recorded (the role record, a ticket, a message or nowhere); and the open work with whether it was listed at the handoff and each item with who now holds it. The signed report gives the handoff, the description with whether it was carried over and the wording found, the revocation item by item, the acceptance and the open work, and counts them, and names an inherited role: a description carried over at the same version or passed on unchanged with no rewrite, a prior holder who keeps any of the role’s permissions, an acceptance not given or not recorded and open work never listed; then a rewrite or revocation after the handoff date, an acceptance recorded only in a message, open items with no holder and a rewrite with no record of what changed, with the fix for each and what would change the verdict. An unstated rewrite, revocation, acceptance or open-work list leaves the handoff unverified. This reads one role passing from holder to holder against the rewrite rule: a task passed between agents is the Agent Handoffs checks such as wever_handoff-acceptance-check, a takeover package for a receiver is wever_handover-check and one half-finished job’s state is wever_job-handoff-pack. Verdict role_handoff_inherited, role_handoff_partial, role_handoff_unverified or role_handoff_rewritten. POST /api/role-handoff-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-scope-boundary-checkAgent Roles: submit a role’s scope boundaries with the work its agent took on: the role with its id, title, the work it covers and the work it excludes; each piece of work with its…Can send data out2 findings
Why it is in this group
- The input addresses people (input.refusal.to): what it sends cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.refusal.to): what the tool writes goes there.
Findings
- Info
The description is 3,460 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
- Warn
The server marks "wever_role-scope-boundary-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.refusal.to): what it sends cannot be taken back.
Description, as the server wrote it Agent Roles: submit a role’s scope boundaries with the work its agent took on: the role with its id, title, the work it covers and the work it excludes; each piece of work with its id, the task, how the records place it against the role (in scope, out of scope or unlisted), how it was handled (done, escalated, refused or handed off) and who approved it; the escalation triggers for out-of-scope requests, each with who it goes to and how fast it must be answered, and the escalation rules as written, read for judgment wording such as when appropriate or if needed; and the refusal path for excluded work (refuse and tell, hand off, escalate or none) with who receives it. The signed report gives the boundaries, every piece of work with its status as in_role, routed, excluded_done, approved_outside, unapproved_outside or unread, every trigger with its receiver and response time and the judgment wording found, and the refusal path, and counts them, and names an open scope: no written scope, excluded work done rather than routed, unlisted work done with no approval, no escalation trigger and no refusal path for excluded work; then triggers with no receiver or no response time, judgment wording in the triggers, approved work the role does not list and a handoff or escalation path with no receiver, with the fix for each and what would change the verdict. Unstated exclusions, work, handling, triggers or refusal path leave the scope unverified. This reads one role’s boundaries against the work its agent took on: a goal’s working scope against what the principal authorized is wever_goal-scope-check, whether escalation triggers are operable is wever_escalation-threshold-check and where a refused or failed task goes on a handoff route is wever_handoff-failure-check. Verdict role_scope_open, role_scope_partial, role_scope_unverified or role_scope_bounded. POST /api/role-scope-boundary-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_rollback-path-checkAgent Recovery: submit the effects a task can leave outside itself (a payment, a message, a booking, an order, a record, an access change or another external write), each with how …1 finding
Why it is in this group
- The description says "cannot be undone".
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_rollback-path-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "cannot be undone".
Description, as the server wrote it Agent Recovery: submit the effects a task can leave outside itself (a payment, a message, a booking, an order, a record, an access change or another external write), each with how it is undone (automatically, by a runbook, by a compensating action, not at all, or it cannot be undone), the reversal as written, how long it stays possible, whether it is marked irreversible before the task runs, whether it waits for a person’s approval, whether the reversal has been run and when it was written down, plus when a task first failed and needed its effects undone. The signed report says for each effect whether it can be reversed, whether the reversal was written before the first failure and which effects are marked irreversible up front, and names every effect with no reversal and no mark, every irreversible effect left unmarked and every reversal never written down, then every irreversible effect run without approval, every reversal never run and every reversal written after the first failure, with the fix for each. This reads the way back for a failed task’s effects: money back from a provider for a failed paid call is wever_refund-path-check, backups and restore steps for data are wever_recovery-readiness-check and what a stop does to in-flight actions is wever_kill-switch-check. Verdict undo_undefined, rollback_gaps, rollback_unverified or undo_defined. POST /api/rollback-path-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_service-terms-checkAgent preflight round 3: read the terms of a service your agent is about to start paying for, from the paying side, before it commits.1 finding
Why it is in this group
- The input carries an amount (input.exposure.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_service-terms-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.exposure.amount) and a currency, so it moves money.
Description, as the server wrote it Agent preflight round 3: read the terms of a service your agent is about to start paying for, from the paying side, before it commits. Input text, fields (price, price_changes, auto_renewal, liability_cap, indemnity, termination, data_use) or both, and an optional exposure (the amount the service can move or charge). The signed report restates each term plainly and flags the one-sided and missing ones: a price that can change with no notice, renewal with no way to cancel, a liability cap below the amount the service can move, liability excluded, one-way indemnity, no termination path for you, termination at will by the provider and broad data use. To verify what a counterparty’s published terms say, use wever_terms-evidence-check. POST /api/service-terms-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-cost-checkTokenized Asset Purchase: submit the full cost of buying and holding a tokenized asset with the cost stated: the stated cost with the headline as a share of the amount and as an am…2 findings
Why it is in this group
- The input carries an amount (input.purchase.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,282 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
- Warn
The server marks "wever_tokenized-cost-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.purchase.amount) and a currency, so it moves money.
Description, as the server wrote it Tokenized Asset Purchase: submit the full cost of buying and holding a tokenized asset with the cost stated: the stated cost with the headline as a share of the amount and as an amount, whether it is once or each year, and its wording, read for no fees, zero commission, all-in or no hidden fees; the purchase with the amount invested, the currency, the years the buyer means to hold and whether the buyer means to sell or redeem; each cost with its id, label, stage (entry, exit, spread, management, carry, custody, network, redemption or other), kind (a percent, a flat amount, a percent each year or a flat amount each year), value and whether the headline counts it; and optionally the points the cost may run over the headline (0.05 when left out). The signed report gives every cost priced on the amount over the years held, exit and redemption costs counted only when the buyer exits, the totals once and per year with each stage’s share, and the headline with the basis it is compared on and the wording found, and counts them, and names an understated cost: a full cost above the headline on its own basis, a cost the headline leaves out and fee-free, hidden-fee or all-in wording the costs contradict; then a headline above the listed costs and yearly costs under a headline with no stated basis, with the fix for each and what would change the verdict. No headline, no amount for percentage costs, no years for yearly costs, no exit or redemption cost, no spread and costs with no headline flag leave the full cost unverified. This reads one purchase’s full cost against its stated cost: charges paired with a fee schedule are wever_fee-evidence-check and charged fees recomputed against a schedule are wever_fee-reconciliation-check. Verdict tokenized_cost_understated, tokenized_cost_partial, tokenized_cost_unverified or tokenized_cost_complete. POST /api/tokenized-cost-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-cancellation-checkAgent Travel: submit the cancellation terms on a booking with what the booking states and any cancellation: the fare rules’ terms with whether the ticket is refundable, what a time…2 findings
Why it is in this group
- The input carries an amount (input.paid.amount) and a currency, so it moves money.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,799 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
- Warn
The server marks "wever_travel-cancellation-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.paid.amount) and a currency, so it moves money.
Description, as the server wrote it Agent Travel: submit the cancellation terms on a booking with what the booking states and any cancellation: the fare rules’ terms with whether the ticket is refundable, what a timely cancellation returns (cash, credit or nothing), the hours before departure by which it must be cancelled, the cancellation fee and currency, the days a credit stays usable and how cancelling part of the trip is treated (prorated, repriced or not allowed); the terms stated on the booking with the cancel-by deadline, refund form, fee, credit expiry, partial treatment and text; the original departure; what was paid; and optionally the cancellation with when it was made, whether only part of the trip was cancelled, the refund form and amount, the fee charged and the credit expiry. The signed report gives the terms, what the booking states, every point with the rules beside the booking and whether the booking states more or less than the rules give, with the cancel-by deadline worked out from the departure, and the cancellation with whether it came inside the deadline, what was owed, what was returned and the credit expiry the rules give, and counts them, and names misstated cancellation terms: a cancel-by deadline later than the rules give, a cash refund stated where the rules return credit or nothing, a fee stated lower than the rules, a credit expiry stated later than the rules and a timely cancellation paid short in amount, form or credit life; then no deadline stated, a credit with no expiry named, no partial treatment stated and terms stated more strictly than the rules, with the fix for each and what would change the verdict. No refund form, no deadline in the rules, no departure, no stated terms or no record of what was paid leaves the terms unverified. This reads one booking’s cancellation terms against its fare rules: a refund claim against a merchant’s policy is wever_refund-eligibility-check, a vendor’s refund terms as a remedy path are wever_vendor-refund-check and a guarantee against its terms is wever_guarantee-evidence-check. Verdict travel_cancellation_misstated, travel_cancellation_partial, travel_cancellation_unverified or travel_cancellation_as_stated. POST /api/travel-cancellation-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_vendor-refund-checkAgent Vendors: submit the refund and cancellation terms for one vendor engagement: the terms as written, read for wording that refuses refunds such as no refunds or all sales final…2 findings
Why it is in this group
- The name "wever_vendor-refund-check" says "refund": a word that usually means money moves, which cannot be taken back.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,252 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
- Warn
The server marks "wever_vendor-refund-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_vendor-refund-check" says "refund": a word that usually means money moves, which cannot be taken back.
Description, as the server wrote it Agent Vendors: submit the refund and cancellation terms for one vendor engagement: the terms as written, read for wording that refuses refunds such as no refunds or all sales final, leaves them wholly to the vendor such as at our sole discretion or pays back in credit such as store credit, the refund window, the conditions a refund depends on, who decides (the buyer, the vendor, the platform, a third party or automatic), how the money comes back, how soon, how to cancel before delivery, the remedy short of a refund, fees taken from a refund and when the terms were stated; and optionally when the spend is planned. The signed report gives the remedy path with the window and the timeline in days and any blocking wording, each part (window, conditions, who decides, how the money comes back, how soon, cancellation) with its status, and counts the parts stated, the conditions and the fees, and names a remedy path that is blocked: refunds refused or a 0 day window, a refund at the vendor’s sole discretion and money back only as credit; then no window or one with no number of days, no conditions, no decider, no method, no timeline, no cancellation path, fees taken from a refund and terms stated after the spend, with the fix for each. This reads the whole remedy path before the spend: a warranty claim against its terms is wever_warranty-check, a refund claim against a policy after purchase is wever_refund-eligibility-check, the refund route for failed paid calls is wever_refund-path-check, a seller’s own checkout refund rule is wever_refund-rule-check and a guarantee against its terms is wever_guarantee-evidence-check. Verdict refund_blocked, refund_partial, refund_unstated or refund_path_clear. POST /api/vendor-refund-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_x402-payment-gatewayPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- The name "wever_x402-payment-gateway" says "payment": a word that usually means money moves, which cannot be taken back.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: ".../x402-payment-gateway. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Check an EOA payment signature and read its Base USDC nonce, balance and chain time. Operating boundary: Supply the payment signature and its terms. Read the signature check, nonce, balance and chain time. An agent uses these observations before settlement; a human reviews the payment terms in the wallet. POST /api/x402-payment-gateway. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current money service follows its published inspection, policy evaluation or private invoice contract. Payment is an API access fee and grants no spending authority.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
Changes things 386Claude Code: ask
edge-alertFree live prediction-market alerts from public Polymarket, Kalshi and Novig APIs.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Free live prediction-market alerts from public Polymarket, Kalshi and Novig APIs. Returns market, side, executable market price, estimated fair price, estimated edge and UTC timestamp. Reuses the thesis fair-value comparison and publishes only estimated edges of 3 percentage points or more. Coverage is bounded to matched pregame sports winner contracts with verified event identity and compatible settlement rules. Each scan names its checked sports, leagues, sources and coverage limits. Prices use a common $1 payout basis and identify their sources. Returns explicit no_alerts when none qualify, or unavailable when live sources cannot be checked. Call with {}. Free on every call, with no x402 charge, wallet or account. Human page: https://weverlabs.com/edge-alerts/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
fair-valueName one prediction market or paste its Kalshi or Polymarket ticker or URL.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Name one prediction market or paste its Kalshi or Polymarket ticker or URL. Plain game names resolve to outcome choices when several contracts match. Returns one existing thesis-model fair-price estimate, its original comparison and timestamps, plus matched source-labelled quotes from Kalshi, Polymarket and Novig when available. Free anonymous access. No wallet or payment is required. POST /api/fair-value. Human example: https://weverlabs.com/fair-value/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_accessibility-evidence-checkDiligenceOps: submit an accessibility statement or conformance claim.1 finding
Why it is in this group
- The input takes a statement or command to run (input.statement), which can change anything it can reach.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_accessibility-evidence-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input takes a statement or command to run (input.statement), which can change anything it can reach.
Description, as the server wrote it DiligenceOps: submit an accessibility statement or conformance claim. The signed report marks the standard and level claimed (WCAG version and A, AA or AAA, Section 508, EN 301 549), the date of the claim, the evaluation method, known limitations and a feedback contact present, missing or ambiguous, with quotes, and flags full conformance claimed beside known limitations, a claim with no method and a claim dated more than a year before the check. POST /api/accessibility-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-disclosure-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_action-threshold-checkAgent Oversight: submit an inventory of agent action classes (name, category, gate, the amount or count a threshold fires above, and the most one action or period can carry).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit an inventory of agent action classes (name, category, gate, the amount or count a threshold fires above, and the most one action or period can carry). The signed report names every consequential class (payments, limit changes, external sends, deletions) with no gate or with a threshold that never fires, the thresholds the material cannot verify, and coverage by category. Controls at the action: approver bands are wever_approval-chain-check and spending policy controls wever_spending-policy-check. Verdict gaps_found, thresholds_unverified or covered. POST /api/action-threshold-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_actor-attribution-checkAgent Accountability: submit an action or audit record design: each kind of action with how its record identifies the actor, the authorization, the permissions held and the time, p…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Accountability: submit an action or audit record design: each kind of action with how its record identifies the actor, the authorization, the permissions held and the time, plus whether permissions are enforced grants or instructions and whether grants keep earlier versions. The signed report names every recorded action that cannot be attributed to a specific agent acting under a specific authorization with specific permissions at that moment, and the missing field in each case. Run completeness is wever_run-record-check. Verdict unattributable_actions, attribution_unverified or attributable. POST /api/actor-attribution-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_adoption-readiness-checkAgent Readiness: submit the rule set an operator plans to hand an agent, by area (spending limits, approval gates, accountability owner, recourse and refund handling, logging), as …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Readiness: submit the rule set an operator plans to hand an agent, by area (spending limits, approval gates, accountability owner, recourse and refund handling, logging), as fields or as the rules in words. The signed report gives each area ready, thin or missing against the situations an agent meets in its first week (a large payment, payments adding up, an unanswered approval, an absent owner, a refund request, a dispute, a wrong payment, a request for what it did and why), naming each one the rules do not cover and the smallest addition that closes it. Spending policy controls are wever_spending-policy-check. Verdict not_ready, ready_with_gaps or ready. POST /api/adoption-readiness-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_agent-allowance-consolePaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...ent-allowance-console. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Preview whether an action fits the supplied scope, budget, usage, expiry and fee limits. Operating boundary: Supply the policy, usage and proposed action. Read the preview and reasons. An agent uses them to prepare the next request; a human adjusts the policy or approves the action through its service. POST /api/agent-allowance-console. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_agent-attention-briefAgent Operations: submit a list of running agents or jobs as reported (name, status, last activity, pending question or blocker, deadline and recent errors where stated) with the t…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Operations: submit a list of running agents or jobs as reported (name, status, last activity, pending question or blocker, deadline and recent errors where stated) with the time of the snapshot. The signed brief triages them in priority order: which need the human first and why (deadline passed, error loop, stopped with an error, blocked on an answer, deadline at risk), which look stalled, which run fine and which finished. Supplied status only: the agents are read from the list and never polled. Verdict human_needed, stalled_or_unclear or all_clear. POST /api/agent-attention-brief. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_agent-budget-guardPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others writeCan send data out3 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.request.callback_url): a request to a URL can carry data to whoever runs it.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...pi/agent-budget-guard. Existing product credentials, signed mandates, and single-use action..."
- Info
The input schema at properties.policy.properties.allowed_callback_origins.items.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.policy.properties.allowed_callback_origins.items.description: "...gin only, with no path, trailing slash, credentials, query or fragment. Matching uses the p..."
- Info
The input schema at properties.request.properties.callback_url.anyOf[0].description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.request.properties.callback_url.anyOf[0].description: "ASCII HTTPS URL without credentials, control characters, backslashes or fra..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Compare a proposed request with the policy, usage and budget values you supply and return an allow-or-deny preview. Operating boundary: Supply policy, usage and request values. Read the preview and reasons. An agent checks the preview before the payment flow; a human sets the budget and approves any exception. POST /api/agent-budget-guard. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current money service follows its published inspection, policy evaluation or private invoice contract. Payment is an API access fee and grants no spending authority.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_agent-card-checkAgent preflight round 2: check an A2A agent card before you rely on the agent.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight round 2: check an A2A agent card before you rely on the agent. Input card as JSON (name, description, url, provider, version, skills, securitySchemes and security or authentication, capabilities) or the same JSON as text. The signed report marks each card element present or missing and checks the card against itself: the card URL and every other published address on HTTPS and a public host, every skill with an id, a name and a description, every authentication requirement backed by a named scheme with the fields its type needs, and capabilities declared as true or false. POST /api/agent-card-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_agent-daily-operator-briefPaid URL: 0.25 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...-daily-operator-brief. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.25 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Free computation: /api/mcp or /api/service-resolver. Read the backend 402 body for x402 terms. Turn supplied service health and queue observations into a prioritized operator brief. Operating boundary: Supply service health and queue observations. Read the ordered priorities and their source observations. Agents prepare the brief; humans use it to choose the next operational task. POST /api/agent-daily-operator-brief. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_agent-identity-checkAgent preflight: check an agent identity record before you deal with the agent.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight: check an agent identity record before you deal with the agent. Input record as JSON (for example an ERC-8004 registration file or a passport excerpt: agent ID, name, endpoints or services, wallet, operator, registrations) or as text. The signed report marks each identity element present or missing and checks the record against itself: the agent ID against its registrations, one wallet across every field that names one, HTTPS endpoints on public host names, EIP-55 wallet checksums and CAIP-10 registry formats. POST /api/agent-identity-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_agent-paid-workflowPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...i/agent-paid-workflow. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Turn a task into an ordered plan of supported computations, with the fields needed for each step. Operating boundary: Supply the task and supported steps. Read the ordered plan and required fields. An agent submits each prepared step through its listed service; a human reviews the sequence and budget first. POST /api/agent-paid-workflow. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_agent-registry-checkAgent Control: submit an agent registry or inventory (each agent with its aliases, owner, purpose, access scope and status) and the deployment, usage, credential or billing lists i…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...and status) and the deployment, usage, credential or billing lists it should cover. The s..."
- Info
The input schema at properties.sources.items.properties.kind.enum[2] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.sources.items.properties.kind.enum[2]: "credentials"
Description, as the server wrote it Agent Control: submit an agent registry or inventory (each agent with its aliases, owner, purpose, access scope and status) and the deployment, usage, credential or billing lists it should cover. The signed report names every agent those lists show with no registry entry, every entry with no named owner, every entry nothing else shows as active or listed retired while a list shows it running, and each field the registry is missing to be authoritative (owner, purpose, access scope, status). Verdict not_authoritative, incomplete or authoritative. POST /api/agent-registry-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_agent-work-order-exchangePaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...t-work-order-exchange. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Validate a task with a local dry run, then prepare the POST request for its supported computation. Operating boundary: Choose a supported computation and supply its task fields. Review validation and the prepared POST request. The agent submits that request to the named endpoint; the human reviews the task and result. POST /api/agent-work-order-exchange. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_agent-work-order-railPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...agent-work-order-rail. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Free computation: /api/mcp or /api/service-resolver. Read the backend 402 body for x402 terms. Prepare a validated DiligenceOps task for a specific Connect recipient. Submission, recipient permission and execution remain separate steps. Operating boundary: Supply the DiligenceOps task and Connect recipient. Read the validated task package. The agent submits it through Connect with the recipient’s permission; the human reviews the requested work and returned result. POST /api/agent-work-order-rail. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_ai-disclosure-checkDiligenceOps: submit content or product material, with the disclosure you want checked or on its own.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit content or product material, with the disclosure you want checked or on its own. The signed report states whether AI involvement is disclosed, where the disclosure sits relative to the content (opening, body or closing, or outside the material), whether its wording names what was generated or assisted (article, images, video, voice, code) and whether it says generated, assisted or both, with any human review clause, hedged wording and assistant wording left in the material quoted. POST /api/ai-disclosure-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-disclosure-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_alert-routeStanding jobs: register one routing rule for the paying wallet.Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Standing jobs: register one routing rule for the paying wallet. Watch fires and standing-job results for this wallet that match the filter also post to callback_url, byte for byte, with their own Idempotency-Key. filter.source_tool is wever_watch-fire or a schedulable tool name; filter.condition reads a field of the delivered report (for example /outcome equals failed, /result/report/verdict equals outside_window, or /observation/value above 120) with equals, not_equals, above, below, at_or_above, at_or_below or exists. Routing never changes the original delivery; it adds one. The rule ends at expires_at, up to 30 days ahead, or after max_deliveries. POST /api/alert-route. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_answer-freshness-checkAgent Operations: submit prior agent answers (each question, the answer given, when, and its basis or source where recorded) with the policy or source change that supersedes part o…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Operations: submit prior agent answers (each question, the answer given, when, and its basis or source where recorded) with the policy or source change that supersedes part of their basis, as the old and new text of each changed section or a stated change description. The signed report flags every answer whose basis the change touches, quotes the superseded basis, and marks each flagged answer stale or unaffected with the reason. Supplied material only: comparing two full policy versions is wever_terms-change-check. Verdict stale_answers, touched_but_current or none_touched. POST /api/answer-freshness-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_api-change-checkDiligenceOps: submit two samples of the same API response (before and after, as JSON or its serialized string).Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a URL (input.endpoint): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it DiligenceOps: submit two samples of the same API response (before and after, as JSON or its serialized string). Following the wever_data-evidence-check field analysis, the signed report lists the fields added, the fields removed, the fields whose JSON type changed and the fields whose value format changed (timestamps, numeric strings, ids, urls and more), by record-relative JSON pointer. POST /api/api-change-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_approval-chain-checkAgent preflight round 3: check your own approval and escalation arrangement before an agent spends under it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight round 3: check your own approval and escalation arrangement before an agent spends under it. Input text, fields (approvers as text or bands of {approver, from, up_to, required}, dual_control, fallback, expiry, currency) or both. The signed report lays the amount bands end to end from zero, states who approves each band and whether it takes two approvers, and names each gap plainly: an amount range no one approves, a single approver with no ceiling, the agent approving itself with no ceiling, no dual-control threshold, no fallback when the approver is unavailable, and approvals that never expire. POST /api/approval-chain-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_approval-durability-checkAgent Control: submit an approval record design: how the approved tool and arguments are stored (exact, summarized, partial or absent), where the record lives, what it references i…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Control: submit an approval record design: how the approved tool and arguments are stored (exact, summarized, partial or absent), where the record lives, what it references instead of holding, its expiry and approval time, and how the call is built at execution, with the longest pause before execution. The signed report names every field whose absence forces reinterpretation at execution (arguments not exact, references to conversation context, a record kept in context, a rebuilt call, no expiry) and whether the stored record can be executed as approved after the pause. Matching an executed call to its approval is wever_approval-replay-check. Verdict reinterpreted_at_execution, durability_unverified or executable_as_approved. POST /api/approval-durability-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_approval-policy-checkAgent Accountability: submit a spend-control design with both momentary approvals (approvers, when an approved payment is checked against standing limits, whether approving changes…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Accountability: submit a spend-control design with both momentary approvals (approvers, when an approved payment is checked against standing limits, whether approving changes a limit) and standing limits set in advance, its conflict rule, its exception path and its execution paths. The signed report says whether approvals and standing limits are distinguished at all, which one wins when a one-off approval exceeds a standing limit, and every path where approving in the moment silently overrides policy set in advance. One action against one approval and policy is wever_policy-precedence-check. Verdict approval_overrides_policy, conflict_unresolved or policy_holds. POST /api/approval-policy-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_approval-record-checkAgent Governance: submit an approval-and-execution design: what the approval record holds (the exact call, its digest, a summary or a reference), where it lives and whether it can …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Governance: submit an approval-and-execution design: what the approval record holds (the exact call, its digest, a summary or a reference), where it lives and whether it can change, each step that carries the call to the executor with what it passes and whether compaction or a rewrite can touch it, and what the executor runs, how it confirms the call matches the approval and what a mismatch does. The signed report names approvals recorded as summaries or kept in conversation history, execution that reconstructs the call instead of running the recorded one, and every point where the executed call can differ from the approved call without detection. The stored form after a long pause is wever_approval-durability-check and one executed call against its approval is wever_approval-replay-check. Verdict divergence_undetected, fidelity_gaps, fidelity_unverified or executes_as_approved. POST /api/approval-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_approval-replay-checkAgent Oversight: submit the approval record (who approved, and where recorded the exact tool and arguments approved or the approval as worded) and the call as executed or as recons…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit the approval record (who approved, and where recorded the exact tool and arguments approved or the approval as worded) and the call as executed or as reconstructed after context compaction. The signed report says whether execution replays exactly what was approved, naming every argument that differs, was dropped or was added, with both values, and whether the tool matches. An approval with no exact arguments reads as unprovable. Tool-call approvals: approved documents against their current version are wever_approval-version-check. Verdict drifted, unprovable or exact. POST /api/approval-replay-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_approval-version-checkAgent Operations: submit one invoice, quote, agreement or order as originally approved, its approval record (who approved it and, where recorded, the version, amount, reference or …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Operations: submit one invoice, quote, agreement or order as originally approved, its approval record (who approved it and, where recorded, the version, amount, reference or SHA-256 it approved) and the current version of the same document. The signed report diffs the two sentence by sentence and returns whether the approval still covers the current version: unchanged, changed only in non-material ways (each listed), or materially changed in amount, parties, scope, dates or terms, with every material change itemized and re-approval required. Approval coverage: contract redlining is wever_contract-diff-check and invoice-to-payment matching is wever_invoice-payment-match. Verdict reapproval_required, non_material_changes or unchanged. POST /api/approval-version-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_atlas-asset-passportLook up a covered tokenized asset by its id or ticker and return its full sourced Asset Passport, including taxonomy stage, six-field rights matrix, claim, backing, access, liquidi…Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Look up a covered tokenized asset by its id or ticker and return its full sourced Asset Passport, including taxonomy stage, six-field rights matrix, claim, backing, access, liquidity, risks and source trail. Reads the versioned Atlas reference dataset, not live market quotes. Each result preserves source dates and labels for provisional or unverified historical display data. POST /api/atlas-asset-passport. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof must bind to the exact JSON arguments and this API path, not /mcp. The adapter forwards only caller-supplied proof and payment headers. Authority is read-only: no dataset or external writes, asset trading, autonomous source fetches, spending authority or publication. API access billing is separate: it records free usage or settles an explicit caller-supplied payment authorization. Read /api/atlas-contract and the backend 402 body for exact schemas and billing terms.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_atlas-issuer-briefAtlas evidence: submit an issuer name as the Tokenization Atlas dataset spells it, for example Paxos Trust Company, NA.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Atlas evidence: submit an issuer name as the Tokenization Atlas dataset spells it, for example Paxos Trust Company, NA. The signed brief lists that issuer’s assets with their taxonomy stages and the six rights fields per asset (issuer involvement, custody and backing, distributions, voting, legal status, redemption), counting disclosed against not disclosed per field and in total. An unknown name returns issuer_not_found with the dataset issuer names that contain it. POST /api/atlas-issuer-brief. Reads the published Tokenization Atlas dataset at https://atlas.weverlabs.com/api/assets once per call, with no dataset writes and no issuer contact. Every report names the dataset version it read, and a field or history the dataset does not hold is reported as not in the published dataset. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/atlas-evidence-contract. Human page: https://atlas.weverlabs.com/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_atlas-market-briefAtlas evidence: call with no arguments.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Atlas evidence: call with no arguments. The signed brief covers the whole board in the published dataset version read: the asset count, the assets and rights counts per stage, and disclosed against not disclosed across the six rights fields, per field and in total, beside the catalog rights coverage the dataset publishes. POST /api/atlas-market-brief. Reads the published Tokenization Atlas dataset at https://atlas.weverlabs.com/api/assets once per call, the same read the Atlas evidence tools make, with no dataset writes and no issuer contact. Every report names the dataset version it read, and a field the dataset does not hold is reported as not in the published dataset. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/atlas-board-contract. Human page: https://atlas.weverlabs.com/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_atlas-redemption-checkAtlas evidence: submit an asset id or ticker.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Atlas evidence: submit an asset id or ticker. The signed report gives the redemption right on record with its detail, passage and review history, the public sources it cites and other sources listing redemption, the recorded redemption route and investor requirements with their supporting sources, unresolved redemption items, and which redemption questions remain not disclosed or not in the published dataset. POST /api/atlas-redemption-check. Reads the published Tokenization Atlas dataset at https://atlas.weverlabs.com/api/assets once per call, with no dataset writes and no issuer contact. Every report names the dataset version it read, and a field or history the dataset does not hold is reported as not in the published dataset. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/atlas-evidence-contract. Human page: https://atlas.weverlabs.com/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_atlas-rights-historyAtlas evidence: submit an asset id or ticker.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Atlas evidence: submit an asset id or ticker. The signed report follows each of the six rights fields across the published dataset versions the server reads, the public API’s current version and the earlier version bundled with the server, naming the version stamps where a field moved between disclosed and not disclosed or its value, detail or sources changed. An asset present only in the current version returns current_version_only with its current state. POST /api/atlas-rights-history. Reads the published Tokenization Atlas dataset at https://atlas.weverlabs.com/api/assets once per call, with no dataset writes and no issuer contact. Every report names the dataset version it read, and a field or history the dataset does not hold is reported as not in the published dataset. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/atlas-evidence-contract. Human page: https://atlas.weverlabs.com/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_atlas-stage-briefAtlas evidence: submit one Atlas stage: wrapper, issuer-sponsored or native.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Atlas evidence: submit one Atlas stage: wrapper, issuer-sponsored or native. The signed brief lists every asset the published dataset classifies at that stage with its classification basis and the six rights fields (issuer involvement, custody and backing, distributions, voting, legal status, redemption), each marked disclosed or not disclosed, with per-field and total counts beside the catalog stage count. A stage with no assets returns a signed brief with an asset count of 0. POST /api/atlas-stage-brief. Reads the published Tokenization Atlas dataset at https://atlas.weverlabs.com/api/assets once per call, the same read the Atlas evidence tools make, with no dataset writes and no issuer contact. Every report names the dataset version it read, and a field the dataset does not hold is reported as not in the published dataset. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/atlas-board-contract. Human page: https://atlas.weverlabs.com/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_audit-evidence-checkDiligenceOps: submit a counterparty’s audit or attestation report excerpt (SOC 2, SOC 1, ISO 27001, PCI DSS or similar), as text, fields (standard, report_type, auditor, period_sta…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a counterparty’s audit or attestation report excerpt (SOC 2, SOC 1, ISO 27001, PCI DSS or similar), as text, fields (standard, report_type, auditor, period_start, period_end, as_of, opinion, exceptions, qualifications, scope) or both, with optional context (engagement_starts, and the summary the vendor gave you). The signed report states what the report covers, the standard and type, the auditor, the period or single date, the opinion, the exceptions and the scope, and names its limits plainly: a point-in-time report presented as period coverage, exceptions left out of the summary, a period that ended before the engagement starts, a qualified opinion, carved-out subservice organizations and controls left to the user entity. POST /api/audit-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_authority-scope-checkAgent Readiness: submit an agent’s access list (systems, the actions it can perform there, access level, data it can reach) and its authority grant (actions it may take, system, li…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The input schema at properties.access.items.properties.via.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.access.items.properties.via.description: "The credential or role."
Description, as the server wrote it Agent Readiness: submit an agent’s access list (systems, the actions it can perform there, access level, data it can reach) and its authority grant (actions it may take, system, limit, approver, data it may use). The signed report names every reachable action and every reachable data set outside the grant, including broad admin, write or read access, every granted action the access cannot perform, and the scoping change that brings each into line. Approver bands are wever_approval-chain-check. Verdict access_exceeds_authority, authority_exceeds_access or aligned. POST /api/authority-scope-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_bazaar-declaration-checkAgent records round 2: check a CDP Bazaar discovery declaration before you publish it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent records round 2: check a CDP Bazaar discovery declaration before you publish it. Input declaration: the 402 body with accepts and extensions.bazaar, a route declaration with amount, resource and bazaar (the shape Wever routes declare) or the bazaar extension alone, as JSON, JSON text or the base64 PAYMENT-REQUIRED header, up to 64 KB. The signed report says whether the declaration would pass the validation CDP applies and names every problem: the discovery envelope, info validated against its own schema, schema patterns that are not valid regular expressions or that use lookahead (the (?![\s\S]) strict-end suffix, with the accepted fix: a plain $ anchor with the line terminators excluded beside it), price, asset and network consistency across the payment terms, and input and output schema completeness. It evaluates the supplied declaration only and fetches nothing. POST /api/bazaar-declaration-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-records-contract. Human page: https://weverlabs.com/reliability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_benchmark-evidence-checkDiligenceOps: submit a benchmark or performance claim and the source text it cites.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a benchmark or performance claim and the source text it cites. The signed report lists every figure in the claim and whether the source states it (same value and unit, with time, size and rate conversions, or a source figure that rounds to it), quotes the source sentence, and names each figure the source does not support beside the source figures in the same unit. POST /api/benchmark-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_boutique-confirm-deliveryConfirm simulated delivery of a shipped order with order_id, bearer order_token and delivery_reference.Can send data out3 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- outboundThe input takes a URL (input.claim.uri): a request to a URL can carry data to whoever runs it.
Findings
- Info
The input schema at properties.claim.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.claim.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
- Info
The input schema at properties.references.items.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.references.items.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
- Warn
The description contains "all other tools", wording about other tools or servers.
What this check looks forMatched text
description: "...e records. Real escrow/dispute caps and all other tools are unchanged."
Description, as the server wrote it Confirm simulated delivery of a shipped order with order_id, bearer order_token and delivery_reference. Reuses escrow-lite signed release and atomically credits the order-scoped simulated boutique treasury once. action=dispute invokes existing dispute-open service and native escrow freeze with claim {reason_code, statement_sha256}; action=dispute-submit-evidence and dispute-status use the existing dispute services in the same simulation store. action=status reads operator state. Disputed orders cannot release. Resolution remains operator-authorized through the existing dispute rail and is not exposed by this bridge. Simulation only, no real funds, shipping verification, wallet ownership assertion or real treasury configuration. The existing order token is a simulation capability, not merchant or buyer authentication. Rejects payment headers and simulation=false. 120 virtual USDC ceiling, 64 orders, 2 MiB, 24-hour expiry. Synthetic test records are isolated from production custody and dispute records. Real escrow/dispute caps and all other tools are unchanged.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_boutique-create-orderSimulation only.No findings
Why it is in this group
- The name "wever_boutique-create-order" says "create": a word for a change.
Description, as the server wrote it Simulation only. Create a boutique order with product_id and buyer_wallet. USD catalog amount maps exactly 1:1 to virtual USDC on the Base x402 pattern. Optional idempotency_key permits exact create retries; retain the returned bearer order_token. Repeat the same product_id and buyer_wallet with action=initiate_payment, simulate_payment, status, or confirm_fulfillment and order_token. Fulfillment confirmation also requires fulfillment_reference. Reuses existing x402 middleware and escrow-lite state machine with isolated synthetic settlement, no network signer or real funds. Simulation ceiling is 120 virtual USDC; real escrow remains capped at 25 USDC and real checkout is disabled. No merchant inventory reservation, shipping, tax, wallet ownership assertion, real payment authorization or actual fulfillment. All payment headers and simulation=false are rejected. Catalog tools remain read-only snapshots. No API fees or allowance consumption. Orders expire after 24 hours; bounded to 64 active simulations.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_boutique-real-money-acceptanceOne owner-operated Rebecca Belle Boutique acceptance order for a dedicated 1.00 USDC test product on Base mainnet.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it One owner-operated Rebecca Belle Boutique acceptance order for a dedicated 1.00 USDC test product on Base mainnet. No catalog items. Creates a bounded order, observes the existing managed escrow, records test shipment and delivery, and independently checks native USDC receipts. This tool never signs, submits or initiates transfers and rejects payment headers. quote is public; other actions require a request-bound X-Wever-Wallet-Proof for /api/boutique-real-money-acceptance and configured buyer/operator. The human executes separately approved funding and release through the unchanged wever_agent-escrow-lite tool. That rail has separate 0.10 USDC API fees after its free allowance and automatic expiry refunds. Test evidence is not customer revenue or actual merchandise fulfillment.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_boutique-record-shipmentRecord a simulated shipment against a funded boutique order using order_id, bearer order_token and tracking {carrier, tracking_number}.1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Findings
- Warn
The description contains "all other tools", wording about other tools or servers.
What this check looks forMatched text
description: "...e records. Real escrow/dispute caps and all other tools are unchanged."
Description, as the server wrote it Record a simulated shipment against a funded boutique order using order_id, bearer order_token and tracking {carrier, tracking_number}. Marks shipped while keeping all virtual USDC in escrow. Exact retries are safe; conflicting tracking is rejected. Simulation only, no real funds, shipping verification, wallet ownership assertion or real treasury configuration. The existing order token is a simulation capability, not merchant or buyer authentication. Rejects payment headers and simulation=false. 120 virtual USDC ceiling, 64 orders, 2 MiB, 24-hour expiry. Synthetic test records are isolated from production custody and dispute records. Real escrow/dispute caps and all other tools are unchanged.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_breach-path-checkAgent Limits: submit a limit set with its breach behavior: for each limit, what hitting it does to the work (stop, escalate, queue or continue), who an escalation goes to and wheth…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Limits: submit a limit set with its breach behavior: for each limit, what hitting it does to the work (stop, escalate, queue or continue), who an escalation goes to and whether that is a named person, a role, a channel or another agent, how long they have to answer, what happens when no one answers (deny, allow or keep waiting), who clears queued work, when the breach path was written down and when the limit was first hit. The signed report follows each breach to its end and names every limit with no written path, every limit the work continues past, every escalation to no one or that goes through on silence, every escalation to an agent, role or channel rather than a named person or with no time to answer, every queue nobody clears, and every path written down after the first breach, with the fix for each. This reads where the work goes at the wall: whether a cap design names a breach rule at all is wever_exposure-cap-check and when an escalation fires and moves to the next owner is wever_escalation-threshold-check. Verdict wall_not_held, breach_path_gaps, breach_path_unverified or breach_paths_defined. POST /api/breach-path-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-limits-contract. Human page: https://weverlabs.com/agent-limits/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_budget-forecast-briefAgent Treasury: submit a current balance with its date, recurring commitments (amount, cadence, next due date), recent spend totals by period, an optional reserve floor and a forec…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Treasury: submit a current balance with its date, recurring commitments (amount, cadence, next due date), recent spend totals by period, an optional reserve floor and a forecast horizon in days. The signed brief projects outflows by week or month, gives the date the balance first falls below the reserve floor (or the balance at the horizon end when it stays above it) and lists every assumption plainly. Arithmetic on the supplied figures only: no market data and no price forecast. Verdict floor_crossed or floor_held. POST /api/budget-forecast-brief. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_cache-staleness-checkAgent Control: submit a caching design for agent context or instructions: each cached element with what it holds (policy, prices, permissions, instructions, tool schemas and others…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Control: submit a caching design for agent context or instructions: each cached element with what it holds (policy, prices, permissions, instructions, tool schemas and others), its TTL and the events that clear it, what changes underneath with how soon each change must take effect, and the decisions made from cached context. The signed report names every cached element that can outlive a change to its source, every decision that can be made from stale context, and the invalidation rule that closes each gap. Agent memory stores are wever_memory-design-check. Verdict outlives_changes, freshness_unverified or invalidated_on_change. POST /api/cache-staleness-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-backlog-checkAgent Capacity: submit an agent’s backlog records across a window: the window with its start and end dates; each backlog item with its id, what it is, when it opened, when it close…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,277 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit an agent’s backlog records across a window: the window with its start and end dates; each backlog item with its id, what it is, when it opened, when it closed and when it last saw activity; optionally the days with no activity after which an open item counts as abandoned in place (30 when left out); and the oldest an open item should be, in days. The signed report gives the window, the backlog open at its start and at its end with the change, the direction (growing, steady or shrinking) and the items opened and closed between them, every item open at the end with its age and its days since last activity, the oldest item waiting with its age and the median age, and every item abandoned in place with its last activity and days idle, and counts them, and names a growing backlog: more open items at the end of the window than at its start, and open items with no activity past the abandonment window; then a backlog held level rather than worked down, and open items older than the age limit, with the fix for each and what would change the verdict. No window, no items, no age limit or open items with no recorded last activity leave the backlog unverified. This reads a backlog’s size and age across a window from its item records: whether waiting items are re-reviewed at their review point is wever_triage-aging-check, how each item left the queue is wever_triage-disposition-check and a report’s queue figures against its records are wever_triage-report-check. Verdict capacity_backlog_growing, capacity_backlog_partial, capacity_backlog_unverified or capacity_backlog_shrinking. POST /api/capacity-backlog-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-burn-checkAgent Capacity: submit a capacity allowance with its recorded use: the allowance with its name, the amount for the period (tasks, tokens, budget or another unit), its unit, the per…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,439 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit a capacity allowance with its recorded use: the allowance with its name, the amount for the period (tasks, tokens, budget or another unit), its unit, the period start and end and any exhaustion date the records state; each use with its id, date, amount, unit and the reason recorded when it ran high; the date the allowance is read at; and optionally the multiple of a typical day above which a day counts as a spike (2 when left out). The signed report gives the allowance with the days in the period and the days elapsed, the burn with what was used, what is left, the share used, the daily rate and the daily rate the rest can carry, the projected exhaustion date worked out from the records with how far it falls before or after the period end beside the date the records state, and every spike day with its multiple of a typical day and whether a reason is recorded, and counts them, and names an overrunning allowance: an allowance used up before its period ends and a daily rate that runs it out before the period end; then spike days with no reason, a projection that lasts only within a tenth of the period past its end, a stated exhaustion date more than 2 days later than the records give and uses in another unit, with the fix for each and what would change the verdict. No amount, no period, no recorded use or no reading date leaves the burn unverified. This reads one period allowance’s pace: a treasury balance projected against a reserve floor is wever_budget-forecast-brief, whether a limit’s remaining room is seen before acting is wever_limit-visibility-check and how one cap is counted across agents is wever_shared-cap-check. Verdict capacity_burn_overrunning, capacity_burn_partial, capacity_burn_unverified or capacity_burn_on_pace. POST /api/capacity-burn-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-concurrency-checkAgent Capacity: submit an agent’s concurrent runs with its concurrency cap: optionally the agent with its id and name; the cap with the most runs allowed at once, what holds it (a …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,518 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit an agent’s concurrent runs with its concurrency cap: optionally the agent with its id and name; the cap with the most runs allowed at once, what holds it (a scheduler, a gateway, the platform, the agent itself or nothing) and the cap as written; each shared resource with its name and whether it takes one run at a time (exclusive) or many (shared); each run with its id, when it started and ended, the resources it holds, its outcome (completed, running, throttled or failed) and why a throttled or failed run stopped (contention, timeout, error or other); optionally the peak concurrency the system recorded; and the moment runs are read at. The signed report gives the cap with whether it is held outside the agent, the peak worked out from every run’s start and end with when it came, the runs at it, the recorded peak and the headroom left, every pair of runs that overlapped on an exclusive resource, and every run stopped for contention, and counts them, and names runs over capacity: more runs at once than the cap, runs overlapping on an exclusive resource and no cap at all; then a cap held only by the agent or by nothing, runs throttled or failed for contention and a recorded peak that differs from the runs, with the fix for each and what would change the verdict. No cap, no runs, open runs with no reading moment or resources with no stated mode leave concurrency unverified. This reads overlapping run times against the concurrency cap: one spending cap drawn by several agents is wever_shared-cap-check, one resource committed twice in collaborators’ recorded decisions is wever_coordination-conflict-check and the layer that enforces a rate ceiling is wever_limit-substrate-check. Verdict capacity_concurrency_over, capacity_concurrency_partial, capacity_concurrency_unverified or capacity_concurrency_within. POST /api/capacity-concurrency-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-headroom-checkAgent Capacity: submit an agent’s spare capacity with its plan: the capacity with the total and its unit; the plan with the spare capacity it keeps as a reserve and the capacity it…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,443 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit an agent’s spare capacity with its plan: the capacity with the total and its unit; the plan with the spare capacity it keeps as a reserve and the capacity it keeps for urgent work; the load with the capacity in use now, the capacity standing or recurring work takes and the headroom the system recorded; each past demand spike with its id, date, peak demand and whether the records show it absorbed; and each urgent request with its id, date, the capacity it needed and how it was served (on time, delayed or refused). The signed report gives the headroom the load leaves with its share of the total beside the recorded headroom, the planned and urgent reserves with what is left after the reserve, the share standing work takes and what it leaves, every spike against the total with its absorption, and every urgent request with how it was served, and counts them, and names headroom gone: a load at or over the total, headroom under the planned reserve, and headroom under the urgent reserve or an urgent request refused; then standing work that alone eats into the reserve, spikes not absorbed or larger than the total, urgent requests served late and a recorded headroom more than 5% of the total above the headroom the load leaves, with the fix for each and what would change the verdict. No total, current load, reserve, urgent reserve, standing load or spike record leaves the headroom unverified. This reads spare capacity against the reserve the plan calls for: whether a limit’s remaining room is shown before acting is wever_limit-visibility-check and a treasury balance against a reserve floor is wever_budget-forecast-brief. Verdict capacity_headroom_none, capacity_headroom_partial, capacity_headroom_unverified or capacity_headroom_adequate. POST /api/capacity-headroom-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-queue-checkAgent Capacity: submit an agent’s work queue with its service target: the queue with its id, name, the longest an item should wait before work starts in hours and the depth at whic…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,460 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit an agent’s work queue with its service target: the queue with its id, name, the longest an item should wait before work starts in hours and the depth at which it counts as full; each depth sample with its date or time, the depth recorded and optionally the arrivals and completions since the previous sample; each item with its id, what it is, when it entered and when work on it started; and the moment the queue is read at. The signed report gives the queue, the depth at the first and last sample with the change, the direction (growing, steady or shrinking) and the arrivals against the completions where every sample records both, every item with its wait before work started, or until the reading moment while it waits, and whether that passes the target, the median and longest wait with the share within target, and the recorded last depth beside the items the records show waiting then, and counts them, and names a backed-up queue: items still waiting past the target wait, a depth that grew by a fifth or more across the window or from empty, and a depth over the limit; then a smaller rise, items that started past the target and a recorded depth that differs from the items, with the fix for each and what would change the verdict. No target wait, fewer than two depth samples, no items or no reading moment for items still waiting leaves the queue unverified. This reads a queue’s depth and waits against its service target: whether waiting items are re-reviewed at their review point is wever_triage-aging-check, the order items are ranked in is wever_triage-priority-check and the uptime and credit terms in an SLA text are wever_sla-evidence-check. Verdict capacity_queue_backed_up, capacity_queue_partial, capacity_queue_unverified or capacity_queue_flowing. POST /api/capacity-queue-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-scaling-checkAgent Capacity: submit the scaling plan for when capacity runs out, with the load now: the plan with each trigger (what fires it, the metric, the threshold, the action and who acts…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,481 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit the scaling plan for when capacity runs out, with the load now: the plan with each trigger (what fires it, the metric, the threshold, the action and who acts), the lead time in days to add capacity once a trigger fires, the maximum scale with its unit, what happens when the maximum is reached (queue, shed, refuse, degrade, escalate or none) with who owns that step, and the plan as written, read for wording such as when needed, as demand grows or scale as required; and the current capacity, load and daily growth in load. The signed report gives every trigger with its threshold, action, owner and any wording that names no threshold, the runway before the load reaches the capacity at the current growth against the lead time with the margin, the maximum with the days until the load reaches it, and the step at the maximum with its owner, and counts them, and names an absent scaling plan: no trigger named, a lead time at or over the runway, no maximum scale and nothing stated for when the maximum is reached; then triggers with no threshold or with judgment wording, triggers with no owner, a step at the maximum with no owner, a runway under twice the lead time and a maximum the load reaches within 90 days, with the fix for each and what would change the verdict. No lead time, no current capacity, load or growth, or no plan text leaves the plan unverified. This reads the plan for adding capacity against the burn it must keep ahead of: where work goes when it hits a spending limit is wever_breach-path-check, whether alerts carry a threshold and an owner is wever_observability-alert-check and when a decision passes to a person is wever_escalation-threshold-check. Verdict capacity_scaling_absent, capacity_scaling_partial, capacity_scaling_unverified or capacity_scaling_ready. POST /api/capacity-scaling-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-throttle-checkAgent Capacity: submit the throttling rules for work that arrives at capacity, with the requests: the rules with each priority class, its rank (1 served first) and what it does at …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,550 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit the throttling rules for work that arrives at capacity, with the requests: the rules with each priority class, its rank (1 served first) and what it does at capacity (admit, defer or refuse), how soon a deferred or refused requester is told in hours, and the rules as written, read for wording such as case by case, best effort or at our discretion; and each request with its id, its class, when it arrived, whether it arrived at capacity, its outcome (admitted, deferred or refused), when work on it started and when the requester was told. The signed report gives the classes in rank order with the discretionary wording found, every request with its rank, its class rule and whether it followed the rule at capacity, every pair where a lower class was admitted at capacity while a higher class that arrived no later waited or was turned away, and every deferred or refused request with when the requester was told, and counts them, and names throttling that is absent: no written priority class, a priority inversion at capacity and a deferred or refused requester never told; then discretionary wording, a request handled against its class rule, notice after the window, a request in an unlisted class and two classes on one rank, with the fix for each and what would change the verdict. No requests, no record of which arrived at capacity, no notice window or no rules text leaves the throttling unverified. This reads admission at capacity against written priority classes: the order queue items are ranked in against severity and impact is wever_triage-priority-check, how each queue item ends with its deferral reason and return date is wever_triage-disposition-check and where work goes at a spending limit is wever_breach-path-check. Verdict capacity_throttle_absent, capacity_throttle_partial, capacity_throttle_unverified or capacity_throttle_ordered. POST /api/capacity-throttle-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_capacity-workload-checkAgent Capacity: submit an agent’s current workload with its stated capacity: optionally the agent with its id and name; the capacity with the most work the agent is stated to hold …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,479 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
Description, as the server wrote it Agent Capacity: submit an agent’s current workload with its stated capacity: optionally the agent with its id and name; the capacity with the most work the agent is stated to hold at once, its unit, who stated it and when; each item of work in hand with its id, what it is, the units of capacity it takes (1 when left out) and its state (active, waiting or blocked); each utilization sample with its date or time, the load in hand then and the utilization the system recorded in percent; and optionally the days over the limit after which an overload counts as sustained (3 when left out). The signed report gives the capacity, the work in hand with its units, its share of the limit and the units held by blocked work, every sample with its load read as a share of the limit beside the recorded figure, and every run of samples over the limit as an episode with its span, its peak and whether it is sustained, and counts them, and names an agent over capacity: work in hand over the stated limit, an overload that lasts the whole window and a sample over the limit recorded at or under 100%; then brief spikes over the limit, a load at 90% or more of the limit, blocked work holding capacity and recorded utilization more than 5 points off the load, with the fix for each and what would change the verdict. No stated limit, no work in hand or no utilization record leaves the workload unverified. This reads the units of work in hand against the stated capacity over time: an agent’s promised hours against its due dates are wever_commitment-deadline-check, where a limit is enforced is wever_limit-substrate-check and running agents triaged from a status snapshot are wever_agent-attention-brief. Verdict capacity_workload_over, capacity_workload_partial, capacity_workload_unverified or capacity_workload_within. POST /api/capacity-workload-check. Reads the load and capacity records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Capacity reads an agent’s own load and capacity records and checks the load against the capacity that exists: work in hand against the stated limit, a queue against its target wait, concurrent runs against the cap, an allowance’s burn against its period, a backlog across a window, spare capacity against the reserve, the scaling plan against the burn, and admission at capacity against written priority rules. Beside it, the Agent Limits checks such as wever_limit-substrate-check read whether a limit sits in code or in prose and how it is scoped, changed and seen, while a capacity check reads the load itself against the capacity; wever_commitment-deadline-check reads promised hours against due dates, while the workload check reads the work in hand against the stated limit over time; and the Agent Triage checks such as wever_triage-priority-check read how queue items are ranked, reviewed and closed, while the queue, backlog and throttle checks read waits, size and admission against capacity. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_catalog-listing-checkAgent Oversight: submit one tool or service listing (name, description, use_when, input_schema, output_schema, price, example input and output).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit one tool or service listing (name, description, use_when, input_schema, output_schema, price, example input and output). The signed report says whether an agent can decide in seconds to call it: use-when line present, input schema complete, output schema present, price stated and example working against the schemas, each missing piece named with its fix. Supplied schemas are read, never compiled. Discovery declarations are wever_bazaar-declaration-check. Verdict not_decidable, decidable_with_gaps or decidable. POST /api/catalog-listing-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_certification-evidence-checkDiligenceOps: submit certification, compliance or license claims (ISO 27001, SOC 2 Type II, PCI DSS, licenses) and the evidence offered for them.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit certification, compliance or license claims (ISO 27001, SOC 2 Type II, PCI DSS, licenses) and the evidence offered for them. The signed report marks each claim supported, contradicted or never mentioned, naming the standard, type or level the evidence states instead and any validity date before the check date, with quotes. POST /api/certification-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_chain-of-custody-checkAgent Handoffs: submit one task that passes through several agents: each step in order with its holder, who it came from and when, who it went to and when, what that holder changed…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Handoffs: submit one task that passes through several agents: each step in order with its holder, who it came from and when, who it went to and when, what that holder changed and where the step comes from (a record written at the time, the holder’s own report or someone’s memory), plus where the task sits now as stated. The signed report lays the steps end to end and names every step that does not link to the one before, every step resting on memory, a last move no step confirms and a stated holder the records do not show, then every holder whose change is unrecorded, every step resting on the holder’s own report and every interval when no one held the task or two holders did, with the fix for each. This reads one task’s custody across several agents: the records of separate handoffs are wever_handoff-record-check, incident claims against a timeline are wever_incident-evidence-check and custody of keys and accounts in an operator handover is wever_handover-check. Verdict custody_unreconstructable, custody_gaps, custody_unverified or custody_traced. POST /api/chain-of-custody-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_change-proof-checkAgent Readiness: submit a coding agent’s change claim (each change with why, files and spec ids) and the evidence (diff summary by file, test results with what each covers and the …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Readiness: submit a coding agent’s change claim (each change with why, files and spec ids) and the evidence (diff summary by file, test results with what each covers and the changes it is cited for, spec references). The signed report says whether the evidence proves the claim: each claimed change with no diff evidence or no passing test that covers it, each test cited for a change it does not cover, each spec item the changes leave open and each diff entry no claim explains. Verdict unproven, partly_proven or proven. POST /api/change-proof-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_changelog-checkDiligenceOps: submit a changelog or release notes (text, or a public HTTPS url) with an optional claimed_version.Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it DiligenceOps: submit a changelog or release notes (text, or a public HTTPS url) with an optional claimed_version. The signed report lists each entry with its version and date, says whether the claimed version has an entry of its own, is only mentioned or is absent, and flags duplicate versions, versions out of listing order, a higher version dated before a lower one, dates after the check date and unreadable dates, each quoted with its offset. POST /api/changelog-check. Reads the supplied text, or one public HTTPS page read once with GET (no redirects, public DNS only, 512 KB): no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_checkout-clarity-checkAgent Operations: submit merchant page or checkout material as pasted (product page text, checkout summary, shipping text, returns text).1 finding
Why it is in this group
- The name "wever_checkout-clarity-check" says "checkout": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_checkout-clarity-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_checkout-clarity-check" says "checkout": a word for a change.
Description, as the server wrote it Agent Operations: submit merchant page or checkout material as pasted (product page text, checkout summary, shipping text, returns text). The signed report says whether an agent can determine, from that material alone, the final price with shipping and tax where stated, the stock or availability, the delivery estimate and the return policy, each marked determinable, ambiguous or absent with the deciding text quoted, plus the question to settle for each open point. Analysis of supplied text: live probing of an endpoint is wever_endpoint-reliability-check. Verdict terms_absent, terms_ambiguous or all_determinable. POST /api/checkout-clarity-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_claim-evidence-checkDiligenceOps: submit a claim (product, listing or diligence claim, as text or a list of parts) and the evidence text or excerpts offered for it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a claim (product, listing or diligence claim, as text or a list of parts) and the evidence text or excerpts offered for it. The signed report marks each part of the claim supported, unsupported or contradicted by the supplied evidence, quoting up to three evidence lines per part with their excerpt and offset, and naming the terms and figures no evidence carries. POST /api/claim-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-closure-checkAgent Commitments: submit the commitments an agent marked done with their closure evidence: each closure with the commitment id, the commitment, the promisee, when it was marked do…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,732 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
Description, as the server wrote it Agent Commitments: submit the commitments an agent marked done with their closure evidence: each closure with the commitment id, the commitment, the promisee, when it was marked done and by whom (the agent, the promisee or another party), whether the deliverable reached the promisee and when, the record of delivery such as a message id, a file link or a ticket, and whether, by whom and when the promisee acknowledged it. The signed report gives every closure with its delivery, its acknowledgement and its status as closed_with_evidence, undelivered, unevidenced, self_closed, acknowledged_by_other, marked_early or unconfirmed, and counts the deliveries with a record and the acknowledgements by the promisee, by another party or by no one, and names closures with no evidence: commitments marked done with nothing delivered, closed with no record of the delivery, and closed on the agent’s own word with no acknowledgement from the promisee; then acknowledgements from someone other than the promisee and commitments marked done before their delivery, with the fix for each. This closes the agent’s own promises with evidence: closing the loop on feedback received is wever_feedback-closure-check, and checking a delegate’s work against its brief before accepting it is wever_delegation-result-check. Verdict closure_unevidenced, closure_evidence_gaps, closure_evidence_unverified or commitments_closed_with_evidence. POST /api/commitment-closure-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-deadline-checkAgent Commitments: submit an agent’s open commitments with their deadlines and the work left: the capacity with the date the plan is read on and the hours a day the agent gives its…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,652 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
Description, as the server wrote it Agent Commitments: submit an agent’s open commitments with their deadlines and the work left: the capacity with the date the plan is read on and the hours a day the agent gives its commitments (6 when left out); and each commitment with its id, the commitment, the promisee, the date it is due and the hours of work left on it. The signed report gives the capacity with the hours needed, every commitment with its working days and hours available by its date and its status as feasible, impossible, stacked, tight, past_due, finished or unconfirmed, and the load by each due date with the commitments due by then, the hours needed and available, the shortfall and the slack, and names an agent that has promised more than it can deliver on time: deadlines the work left cannot fit on their own, and stacks of commitments that fit one by one but not together by some date; then stacks with under 15% of their hours to spare and commitments already past their date with work left, with the fix for each. This tests the agent’s own deadlines against its own remaining work: the dates and obligations in one document are wever_deadline-check, and one resource committed twice by collaborators is wever_coordination-conflict-check. Verdict deadlines_overcommitted, deadline_feasibility_gaps, deadline_feasibility_unverified or deadlines_feasible. POST /api/commitment-deadline-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-renegotiation-checkAgent Commitments: submit the changes to a commitment’s terms and the edits to its record: each change with its id, the commitment, the term changed (the deadline, the deliverable,…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,085 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
Description, as the server wrote it Agent Commitments: submit the changes to a commitment’s terms and the edits to its record: each change with its id, the commitment, the term changed (the deadline, the deliverable, the scope, the owner, the price or the done test), the term as first agreed and as changed, the reason given to the promisee, the original deadline, whether it was put to the promisee as a proposal rather than announced, when it was proposed, whether and when the promisee agreed, and whether the original terms stay in the record beside the revised ones; and each edit to a commitment record with its id, the commitment, the field edited, when, and the id of the change it carries out. The signed report gives every change with whether it was proposed and agreed before the original deadline and its status as agreed_in_time, announced, unagreed, proposed_late, original_overwritten, agreed_late, reason_missing or unconfirmed, every edit with its status as carries_agreed_change, quiet_edit or unconfirmed, and counts the changes proposed, agreed and in time and the quiet edits, and names terms changed unilaterally: changes announced to the promisee rather than proposed, changes standing without the promisee’s agreement, changes first raised after the original deadline, and edits to a commitment record no agreed change explains; then original terms overwritten, agreement reached only after the deadline and changes with no reason, with the fix for each. This reads changes to the agent’s own promises: a counteroffer read against the original offer and a mandate is wever_negotiation-counteroffer-check, and the record of offers in a deal is wever_negotiation-record-check. Verdict terms_changed_unilaterally, renegotiation_gaps, renegotiation_unverified or renegotiation_honest. POST /api/commitment-renegotiation-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-report-checkAgent Commitments: submit the kept-promises report written for the principal, with the record behind it: its title, its audience, the period it covers and its text, read for wordin…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,848 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
Description, as the server wrote it Agent Commitments: submit the kept-promises report written for the principal, with the record behind it: its title, its audience, the period it covers and its text, read for wording such as every promise kept or 100% on time; for the commitments made, kept, late, renegotiated and broken, whether the report states them, how many it names and whether each links to the record behind its status; and optionally the record of commitments made, kept, late, renegotiated and broken in the period. The signed report gives the report with any perfect-record wording, each status with whether it is stated, its count and whether it is backed by the record, and the record with the commitments still open, and names a report that misstates the record: late or broken commitments left out or stated as none against the record, late or broken commitments named below the record’s count, more kept than the record shows, and statuses with no link to the record behind them; then commitments made, kept or renegotiated left out, other counts off the record’s and perfect-record wording while the record holds late or broken commitments, with the fix for each. This reports on the agent’s own promises against their record: a reputation summary built from a track record is wever_reputation-report-check, one reputation claim is wever_reputation-claim-check, and a group’s report to its principal is wever_coordination-report-check. Verdict commitment_record_misreported, commitment_report_gaps, commitment_report_unverified or commitment_report_sound. POST /api/commitment-report-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-slippage-checkAgent Commitments: submit an agent’s commitments with their progress and any warnings given: the date they are read on; each commitment with its id, the commitment, the promisee, w…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,011 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
Description, as the server wrote it Agent Commitments: submit an agent’s commitments with their progress and any warnings given: the date they are read on; each commitment with its id, the commitment, the promisee, when work started, the date it is due, the share done in percent, when it was delivered if it was, when the promisee was warned it may slip and what that warning measured drift against (the work remaining, the time elapsed or the date passing); and optionally the days of notice a warning gives before the due date (2 when left out). The signed report projects each open commitment’s finish from its pace so far, the share of work left at the share done per day since work started, and gives every commitment with its progress, its projected finish, whether it runs late, its warning with the days of notice it gave and its status as on_track, delivered_on_time, flagged_early, slip_unflagged, missed_unflagged, warned_after_due, short_notice, measured_by_calendar or unconfirmed, and counts the commitments projected late and flagged in time, and names slippage left unflagged: commitments projected to finish late with no warning, deadlines missed with no warning, and warnings given only after the date passed; then warnings with less than the stated notice and warnings that measured drift by the calendar rather than the work remaining, with the fix for each. This reads the agent’s own early warning on its own promises: when an operator’s escalation policy fires is wever_escalation-threshold-check, and whether a delegate reports failures to its principal as they happen is wever_delegation-reporting-check. Verdict slippage_unflagged, slippage_warning_gaps, slippage_warning_unverified or slippage_flagged_early. POST /api/commitment-slippage-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-specificity-checkAgent Commitments: submit commitment records to read for whether each can be verified: each commitment with its id, the commitment as written, read for hedges such as try to, hopef…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,828 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
Description, as the server wrote it Agent Commitments: submit commitment records to read for whether each can be verified: each commitment with its id, the commitment as written, read for hedges such as try to, hopefully or if possible and for timing such as soon or when possible, the deliverable, read for activity wording such as look into or help with, the owner by name, the date it is due, the event it is due on such as when the contract is signed, and how to tell done from not done, read for wording such as when it looks good. The signed report gives every commitment with its parts, the vague wording read from each and its status as specific, deliverable_unclear, owner_unnamed, timing_missing, done_test_missing, timing_vague, done_test_vague, hedged or unconfirmed, counts the specific, owned, dated and tested commitments, and lists every vague word found, and names commitments that cannot be verified: a deliverable that is missing or worded as activity, no owner or an owner named only as a group or a placeholder such as the team or TBD, no date or event, and no way to tell done from not done; then timing no one can check, a done test left to opinion and a hedged commitment, with the fix for each. This reads the agent’s own commitments as written: missing terms and acceptance tests in a proposed deal are wever_negotiation-terms-check, and the accountable person per outcome is wever_ownership-map-brief. Verdict commitments_unverifiable, commitment_specificity_gaps, commitment_specificity_unverified or commitments_specific. POST /api/commitment-specificity-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_commitment-tracking-checkAgent Commitments: submit the open-commitment ledger an agent works from with the commitments it made: optionally the ledger with its owner and the date it was read; each ledger en…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,866 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
Description, as the server wrote it Agent Commitments: submit the open-commitment ledger an agent works from with the commitments it made: optionally the ledger with its owner and the date it was read; each ledger entry with its id, the id of the commitment it tracks, the commitment as the entry states it, the promisee, its status (open, in progress, blocked, done or cancelled), when it was last updated and, for a cancelled entry, whether the promisee released the agent from it; each commitment made with its id, the commitment, the promisee and whether it is still owed; and optionally how many days an open status may go without an update (14 when left out). The signed report gives the ledger, every entry with its age and its status as current, stale, dropped_silently, released, duplicate, orphaned, contradicts_record or unconfirmed, and every commitment made with its status as on_ledger, missing, closed or unconfirmed, and names a ledger the agent cannot work from: open commitments missing from it and entries cancelled with no release from the promisee; then statuses that disagree with the record, open statuses not updated within the interval, two entries for one commitment and entries tied to no commitment made or no promisee, with the fix for each. This reads the agent’s own ledger: a group’s joint progress against its plan is wever_coordination-progress-check, and whether a delegate owes its principal reports is wever_delegation-reporting-check. Verdict commitment_ledger_unreliable, commitment_tracking_gaps, commitment_tracking_unverified or commitments_tracked. POST /api/commitment-tracking-check. Reads only the recorded promises and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no promise is made, changed or closed and no promisee is contacted. Agent Commitments reads the promises an agent makes itself, from capture to closure: it is not delegation (assigning work and authority to another agent, the wever_delegation checks; a commitment is the agent’s own promise, whoever ends up doing the work), not coordination (how agents organize shared work, the wever_coordination checks), not reputation (the durable standing a track record earns, the wever_reputation checks; commitments are the individual promises that record is built from), not the loop on feedback received (wever_feedback-closure-check; this family closes the agent’s own promises) and not who answers for an outcome (wever_ownership-map-brief; this family tracks the promise itself). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-evidence-checkAgent Consent: submit the consent record behind a grant: its id, who gave the consent, how their identity is established (an account login, a verified ID, a signature, an email or …2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 4,121 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
Description, as the server wrote it Agent Consent: submit the consent record behind a grant: its id, who gave the consent, how their identity is established (an account login, a verified ID, a signature, an email or phone confirmation, unverified or none), when they gave it, how it was captured (an explicit opt-in, a signed form, a recorded or unrecorded verbal agreement, a pre-ticked box, a default left on, a line in the terms, inferred, or other), what the principal was shown with the notice text, its version and when it was shown, where the record is kept, the system or person that captured it and the record’s own wording, read for wording that assumes consent such as by continuing you agree, implied consent or unless you opt out. The signed report gives the record with any wording that assumes consent, what was shown and whether it came before the consent, each part (who granted, when, how captured, what was shown, where kept) with its status, and counts the parts stated, and names evidence missing the principal’s agreement: no grantor, no time, no capture method, capture by a pre-ticked box, a default left on, a line in the terms or consent inferred from conduct, wording that assumes consent and a notice shown after the consent; then a notice not kept, no notice version, no time shown, an unrecorded verbal consent, an unverified grantor and no capturer, with the fix for each. A record with no notice, no identity method or no reference to its source is unverifiable. This reads the evidence that one principal agreed: whether a principal agreed to be remembered by a memory system is wever_memory-consent-check, privacy claims against a policy are wever_privacy-evidence-check, a delegation or mandate document’s terms are wever_delegation-check and how a negotiation mandate’s authorization is recorded is wever_negotiation-mandate-check. Verdict consent_evidence_missing, consent_evidence_partial, consent_evidence_unverifiable or consent_evidenced. POST /api/consent-evidence-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-expiry-checkAgent Consent: submit the grant’s term and the planned use: the grant with its id, when it was given, when it ends, how long it runs (such as 12 months or until withdrawn), whether…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 3,809 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
Description, as the server wrote it Agent Consent: submit the grant’s term and the planned use: the grant with its id, when it was given, when it ends, how long it runs (such as 12 months or until withdrawn), whether it runs until the principal withdraws it and its wording, read for an open-ended term such as as long as necessary or indefinitely; the use with when it starts, when it ends and whether it is ongoing with no end; and optionally the moment to read the grant at (the use start when left out) and the days before the end that count as expiring (30 when left out). The signed report gives the grant with its end and where the end comes from (the stated date, the date given plus the term in days, or a rule that it runs until withdrawn), the use, and a timeline with the moment read, the days left and the days the use runs past the end, and names a grant that does not last the use: a grant that ended before the use starts or before the moment read, and a use that finishes after the grant ends; then an ongoing use with no end against a grant that ends, an end inside the warning window, a term that names no length and an open-ended term, with the fix for each. Months count at 30 days and years at 365. This reads the end of one grant against the use: a delegation’s end and whether it is checked before each use is wever_delegation-expiry-check, a trading or treasury mandate’s current state is wever_mandate-status, how long a memory system keeps what it remembers is wever_memory-retention-check and a written data retention policy is wever_data-retention-check. Verdict consent_expired, consent_expiring, consent_expiry_unknown or consent_current. POST /api/consent-expiry-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-purpose-checkAgent Consent: submit the purposes the grant was given for and the planned use: the grant with its id, its purposes and the purposes it rules out, each one of service delivery, acc…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 4,009 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
Description, as the server wrote it Agent Consent: submit the purposes the grant was given for and the planned use: the grant with its id, its purposes and the purposes it rules out, each one of service delivery, account management, customer support, security, legal compliance, analytics, personalization, product improvement, research, marketing, advertising, model training, profiling, data sale or other, and the purpose as the principal was told it, read for wording that stretches it such as related purposes, to improve our services or any purpose; and the use with its main purpose, the other purposes it serves and its wording, read for purposes it does not declare: training a model, advertising, profiling, selling the data and marketing messages. The signed report gives the grant with any stretching wording, the use with any undeclared wording, every purpose the use serves with whether it is declared and its status as aligned, excluded, catch_all, outside or unread, and counts the purposes declared, undeclared, aligned, outside, excluded and reached by stretching wording, and names a use whose purpose the principal did not agree to: a declared purpose the grant does not name, a purpose the grant rules out and an undeclared purpose the grant does not name; then a purpose covered only by stretching wording and an undeclared purpose the grant does name, with the fix for each. This reads one use against the purposes the principal agreed to: whether one sharing event’s new purpose needs fresh consent under the owner’s sharing rules is wever_context-sharing-check, broad data use in a service’s terms is wever_service-terms-check and what a privacy policy says about collection, sharing and retention is wever_privacy-evidence-check. Verdict consent_purpose_mismatch, consent_purpose_stretched, consent_purpose_unclear or consent_purpose_aligned. POST /api/consent-purpose-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_consent-renewal-checkAgent Consent: submit the grant’s renewal terms, the changes since it was given and the moment the use continues: the grant with its id, when it was given, when the principal last …2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
- Info
The description is 4,001 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
Description, as the server wrote it Agent Consent: submit the grant’s renewal terms, the changes since it was given and the moment the use continues: the grant with its id, when it was given, when the principal last confirmed it, how often it must be confirmed again (such as 12 months), how it renews (by the principal’s confirmation, by itself, or not at all) and when the agent last asked the principal to confirm; each change since the grant with its id, when it happened, its kind (a new purpose, new data, a new recipient, a new actor, new terms or a wider use), the change as written and when the principal agreed to it; and optionally the days before the due date that count as pending (30 when left out). The signed report gives the grant with its interval in days, every change with its status as confirmed, unconfirmed or after_read, and a timeline with the due date and the days to it, with counts, and names fresh consent needed before the use continues: a confirmation date already passed and a change the principal has not agreed to by the moment the use continues; then a grant that renews by itself, a due date inside the warning window, a confirmation request still unanswered and an interval with no length, with the fix for each. The due date is the last confirmation, or the grant, plus the interval read in days, months at 30 days. This reads re-confirmation and the changes that need fresh consent: a payout change needing the counterparty’s fresh approval is wever_payout-change-check, one sharing event needing fresh consent under the owner’s rules is wever_context-sharing-check, how a delegation renews when it lapses is wever_delegation-expiry-check and amending a trading or treasury mandate is wever_mandate-amend. Verdict consent_renewal_overdue, consent_renewal_pending, consent_renewal_terms_unknown or consent_renewal_current. POST /api/consent-renewal-check. Reads only the recorded consent material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no principal is contacted, no consent is requested, recorded, renewed or withdrawn and no data is used, moved or deleted. Agent Consent reads the grant the principal gave before the agent uses data or acts on it: whether the grant covers this use, whether it is still in force, whether it was withdrawn, whether the record proves it, what it was given for, whether the data may pass on and whether fresh consent is due. It is not the agent’s authority to act (wever_authority-scope-check reads an agent’s access list against the authority grant its operator gave it; a consent check reads the grant the principal gave over their own data: authority is not consent, and consent is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-amend, wever_mandate-revoke and wever_mandate-draft, issue and change what an agent may trade or spend for a principal; a consent check reads what the principal agreed may be done with their data and changes nothing), not a data sample (wever_data-evidence-check reads a dataset sample’s fields and types; a consent check reads the permission behind using the data, not the data itself), not terms against a purchase (wever_terms-match-check reads purchases already made against their governing terms; a consent grant is the principal’s permission, not a seller’s terms), not a promise (the wever_commitment checks read what the agent promised; consent is what the principal granted, and an agent’s promise never stands in for it) and not an objective (the wever_goal checks read the objective the work serves; a consent check reads whether the principal’s grant allows the work at all). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_context-sharing-checkAgent Operations: submit one proposed sharing event (the context items, who would receive them and in what setting and for what purpose), the owner’s stated sharing rules as struct…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Operations: submit one proposed sharing event (the context items, who would receive them and in what setting and for what purpose), the owner’s stated sharing rules as structured rules (refuse, require consent or allow, by category, item, recipient, setting and purpose) and any prior consent records, with the event date. The signed report says whether an existing consent covers the event, whether fresh consent is needed first and exactly what changed to need it (new participant, new setting, wider scope, new purpose, or an expired or revoked consent), or whether a stated rule refuses it. Consent logic on supplied records only. Verdict refused_by_rule, fresh_consent_required or covered. POST /api/context-sharing-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_continuity-checkAgent preflight: check an operator’s continuity arrangements before you rely on the operator.1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The input schema at properties.fields.properties.key_custody.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.fields.properties.key_custody.description: "Who holds the signing keys and credentials, and any backup or recovery for them."
Description, as the server wrote it Agent preflight: check an operator’s continuity arrangements before you rely on the operator. Input text, fields (key_custody, successor, standing_jobs, subscriptions, escrowed_work, records_export) or both. The signed report marks each area covered, partial or not covered, with quotes, and names the gaps plainly: a single point of failure in key custody, no successor, no plan for standing jobs, subscriptions or escrowed work, and no export path for records. POST /api/continuity-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_contract-briefDiligenceOps: submit one contract excerpt.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit one contract excerpt. The signed brief states the parties with their roles, the term with its effective and end dates and length, renewal with its kind, period and notice, termination with its grounds, notice and cure periods and fees, and payment with its amounts, due terms, frequency and late charges, each quoted, and lists what the excerpt does not state. POST /api/contract-brief. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-boundary-checkAgent Coordination: submit how one job is divided between the participating agents: each part of the job with its name, the other names assignments use for it, whether it is worked…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Coordination: submit how one job is divided between the participating agents: each part of the job with its name, the other names assignments use for it, whether it is worked jointly by design and the agent that leads it, and each agent’s assignment with the parts it is told to do, the parts it is told to leave alone and the assignment as written. The signed report gives every part with the agents whose assignments cover it (owned, overlapping, jointly worked with or without a lead, or unowned), every assignment with the parts it covers, read from its list or else from the parts its statement names, and its wording (bounded, hedged or open-ended), and names territory that collides: parts two agents are told to do, parts in nobody’s assignment and assignments stated so loosely, with anything else, as needed or etc., that overlap is inevitable; then edges stated as help or support, assignments naming parts not on the job’s list, jointly worked parts with no lead and agents with no stated territory, with the fix for each. The access that travels with one handed-off task is wever_handoff-scope-check and the tasks one delegate may do for one principal are wever_delegation-scope-check. Verdict territory_collides, territory_gaps, territory_unverified or territory_clear. POST /api/coordination-boundary-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-conflict-checkAgent Coordination: submit two or more collaborators’ recorded actions and decisions on the same job: each with its id, the agent, its kind (a commitment of a resource, an output, …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Coordination: submit two or more collaborators’ recorded actions and decisions on the same job: each with its id, the agent, its kind (a commitment of a resource, an output, a decision, a reversal or other), the resource it commits or the subject it is about, what it says or decides, when it was taken, the action it reverses, whether an output shipped outside the group, the agents told about it and whether it was withdrawn; optionally the agents on the job; and the rule that settles a collision (the lead decides, the principal decides, the first claim wins, the agents vote, or no rule) with who decides. The signed report counts the live, reversed and withdrawn actions, lists every collision, and names direct collisions: the same resource committed twice, contradictory outputs shipped and a decision one agent reversed without telling the others; then decisions and unshipped outputs that diverge on one subject, no resolution rule and a lead or principal rule with no one to decide, with the fix for each. Collaborators here share one goal: parties with opposing interests stuck on terms are wever_negotiation-deadlock-check, and knowledge entries that contradict each other are wever_knowledge-conflict-check. Verdict collaborators_collide, collision_gaps, collisions_unverified or no_collisions. POST /api/coordination-conflict-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-plan-checkAgent Coordination: submit a multi-agent work plan as recorded: optionally the agents in it, each with its id and the role it holds, and each task with its id, its title, the agent…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Coordination: submit a multi-agent work plan as recorded: optionally the agents in it, each with its id and the role it holds, and each task with its id, its title, the agents that own it, the outputs or deliverables it produces (the task itself when none is named), the outputs of other tasks it needs and the tasks it waits for. The signed report lists every task with its owners, the tasks it waits on and the inputs nobody produces, every deliverable with the agents that own it (single owner, contested or unowned) and every circular wait, and names a plan that cannot run as written: tasks with no owner, deliverables two agents own, circular dependencies and tasks that need an output no task in the plan produces; then dependencies on tasks not in the plan, owners not among the listed agents, listed agents with no task and agents with no role, with the fix for each. This reads how several collaborating agents divide one job: one principal handing work to one agent is wever_delegation-scope-check, a chain of delegations is wever_delegation-chain-check and one handoff record between two parties is wever_handoff-record-check. Verdict plan_unworkable, plan_gaps, plan_unverified or plan_workable. POST /api/coordination-plan-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-progress-checkAgent Coordination: submit a joint progress report from several agents with the plan it reports against: each plan task with the agent the plan gives it to; each progress claim wit…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Coordination: submit a joint progress report from several agents with the plan it reports against: each plan task with the agent the plan gives it to; each progress claim with the task, the agent making it, its status (done, in progress, blocked or not started) and the evidence behind it (a link, a commit, a test run, a file or a record id); and each blocker with its id, the task it blocks, what it is and who is clearing it. The signed report gives every plan task with the agents claiming it and its status (done, in progress, blocked, not started, claimed done twice, disputed or unclaimed), counts the claims, the done claims with evidence and the claims off the plan, lists every blocker as owned or unowned, and names progress that is misreported: tasks claimed done by two agents, plan tasks nobody claims and blockers with no owner; then done claims with no evidence attached, tasks agents report differently, claims off the plan, claims by an agent other than the plan’s owner and tasks reported blocked with no blocker, with the fix for each. A delegate’s duty to report to one principal is wever_delegation-reporting-check, partly done work in one task system is wever_partial-state-check and one run’s trace is wever_observability-trace-check. Verdict progress_misreported, progress_gaps, progress_unverified or progress_accounted. POST /api/coordination-progress-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-report-checkAgent Coordination: submit one report a group of agents writes for its principal, with the record behind it: its title, its audience, its headline as written and optionally its tex…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Coordination: submit one report a group of agents writes for its principal, with the record behind it: its title, its audience, its headline as written and optionally its text; whether it states the collisions between agents, the tasks no agent owns, the open blockers and the unresolved disagreements, where each sits (beside the headline, in a later section, in an appendix or in a separate document) and how many it names; and optionally the record of each, as found in the plan, state, action and progress records. The signed report gives the headline with any frictionless wording, the disclosure of each kind of friction and the record, and names friction smoothed into a clean summary: collisions, unowned tasks, blockers or disagreements left out or stated as none while the record holds some; then counts below the record’s, disclosures in an appendix or a separate document, and wording such as on track, smooth or fully aligned while the record holds friction, with the fix for each. A delegate’s standing duty to report is wever_delegation-reporting-check, a knowledge report is wever_knowledge-report-check and an observability summary is wever_observability-report-check. Verdict friction_smoothed_over, friction_gaps, friction_unverified or friction_disclosed. POST /api/coordination-report-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-roster-checkAgent Coordination: submit the roster of agents taking part in one job: each agent with its id, its name, a verifiable identity (a key id, a DID, a service account or a registry en…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Coordination: submit the roster of agents taking part in one job: each agent with its id, its name, a verifiable identity (a key id, a DID, a service account or a registry entry), its role and what it may read, write and approve; optionally the roles the job needs with whether each is required; and optionally every area that needs an approval, such as merges, spending or publishing. The signed report lists every agent as identified, unidentified or unnamed with its rights, every role with the agents holding it and every approval area with its approvers, and names rights no one can be held to: unnamed or unidentifiable agents holding write or approval rights, roles with no agent in them and one agent holding every approval; then approval areas nobody may approve, areas approved only by the agent that writes them, reading agents with no identity, agents with no role and roles the job does not list, with the fix for each. This reads the participants in one multi-agent job: whether an organisation keeps one current list of all its agents is wever_fleet-roster-check and whether one registry covers every deployed agent is wever_agent-registry-check. Verdict rights_unaccountable, rights_gaps, rights_unverified or rights_accountable. POST /api/coordination-roster-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_coordination-state-checkAgent Coordination: submit how a group of agents shares state: each shared record (a board, a file, a database, a document, a message thread, one agent’s own context or other) with…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Coordination: submit how a group of agents shares state: each shared record (a board, a file, a database, a document, a message thread, one agent’s own context or other) with its id, its name, its owner, the agents that write it and read it, the rule that settles two writes (one writer only, a lock, a versioned merge, the owner decides, the last write wins or no rule), what it keeps (a full change log, snapshots, the latest value only or nothing) and where a joining agent finds it; and optionally the agents in the group. The signed report lists every record with its writers, rule, history and whether a joining agent can rebuild it, counts the records with several, one or no writers and each rule, and names shared state that is unsafe: records with no named owner or writer, records several agents can overwrite with no conflict rule, and state a joining agent cannot reconstruct because it lives in one agent’s context, keeps nothing or has no location; then writers with no owner, last write wins, latest-value-only history and writers outside the group, with the fix for each. One agent’s view against the system record after an interruption is wever_state-reconciliation-check and partly done work in one task system is wever_partial-state-check. Verdict shared_state_unsafe, shared_state_gaps, shared_state_unverified or shared_state_sound. POST /api/coordination-state-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-coordination-contract. Human page: https://weverlabs.com/agent-coordination/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_counterparty-briefWever Briefs: one-call diligence brief on an agent or service.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Wever Briefs: one-call diligence brief on an agent or service. Input counterparty: a Wever agent ID, mailbox ID or registered name, a service or operator name, a domain or a URL. Combines its free agent directory listing where one exists, its Wever x402 index entry, Agent Trust Check facts on file and an evidence summary. Each source states found, nothing found, unavailable or not run, so the brief names exactly which sources were checked and which had nothing. POST /api/counterparty-brief. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns one brief with an Ed25519 signed receipt. Contract: /api/briefs-contract. Human page: https://weverlabs.com/briefs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_counterparty-onboarding-checkCommerce Assurance: submit a counterparty onboarding packet: legal or agent identity, contact, wallet or bank payment route, terms acceptance and compliance fields as supplied, wit…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Commerce Assurance: submit a counterparty onboarding packet: legal or agent identity, contact, wallet or bank payment route, terms acceptance and compliance fields as supplied, with optional requirements (terms version, compliance fields). The signed report checks completeness and internal consistency: email, phone and website formats, the wallet format or the IBAN, SWIFT/BIC and routing formats, the account holder against the legal name, the terms version and date, and compliance dates and open statuses. Verdict complete, incomplete or inconsistent. Research on the counterparty itself is wever_counterparty-brief. POST /api/counterparty-onboarding-check. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_credential-requirements-checkAgent Readiness: submit one tool or service listing and its purchase flow in order (each step with its words, the credentials, accounts or approvals it asks for, and the step where…3 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The title mentions "Credential".
What this check looks forMatched text
title: "Credential requirements check"
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...in order (each step with its words, the credentials, accounts or approvals it asks for, and..."
- Info
The input schema at properties.flow.items.properties.requires.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.flow.items.properties.requires.description: "Credentials, accounts or approvals first asked for..."
Description, as the server wrote it Agent Readiness: submit one tool or service listing and its purchase flow in order (each step with its words, the credentials, accounts or approvals it asks for, and the step where payment happens). The signed report says whether an agent learns every credential requirement before the payment step, names each API key, token, account, authorization, identity check, approval or plan first disclosed at or after payment, and gives the disclosure point the listing should move each one to. Listing decidability is wever_catalog-listing-check. Verdict disclosed_late, requirements_unnamed, disclosed_before_payment or no_requirements_found. POST /api/credential-requirements-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_data-evidence-checkDiligenceOps: inspect a submitted JSON sample and return a signed field-by-field verification report.Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.endpoint): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it DiligenceOps: inspect a submitted JSON sample and return a signed field-by-field verification report. Checks required fields, empty values, expected and consistent types, timestamp parsing and age, source/provider attribution, shape, record count and detectable duplicate keys. Maximum payload 256 KB (262144 UTF-8 bytes). required_fields leads the report; dataset_type selects defaults only when required_fields is absent. endpoint is recorded as supplied without fetching. Facts describe the submitted sample, not seller ratings or underlying data truth. POST /api/data-evidence-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and /api/data-evidence-check. Read /api/data-evidence-check-contract for the field and freshness rules, receipt verification and public key. Human page: https://weverlabs.com/data-evidence-check/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_data-retention-checkDiligenceOps: submit a data retention and deletion policy.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a data retention and deletion policy. The signed report lists each data class the policy names (account data, payment records, logs, backups, communications, analytics and more) with its retention period, a period stated only as "as long as necessary", or none, and marks the deletion method, the deletion timeline after a request and stated exceptions present, missing or ambiguous, with quotes. POST /api/data-retention-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-disclosure-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-coverage-checkAgent Datasets: submit a dataset’s coverage and representativeness claims with its stated population: the claim with the population it is said to represent, each segment with its d…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,909 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s coverage and representativeness claims with its stated population: the claim with the population it is said to represent, each segment with its dimension (such as region or age band), its value and its share of the population, the places it claims to cover, the period it claims and its wording, read for representative, nationally representative, balanced or unbiased; the records per segment with each dimension, value and count; the records per class label; the records per place; the first and last records with any periods that hold no records; and optionally the points a segment may sit from its population share (5 when left out) and the smallest class as a share of the largest (10 percent when left out). The signed report gives every claimed segment with its population share, its records, its share of the dataset, the points it sits off and its status as within, drifting, skewed, missing or unread, the segments the data holds that the claim leaves out, the classes with the smallest as a share of the largest, every claimed place as covered, missing or unread, the time with the claimed period beside the first and last records and the gaps, and counts them, and names a skewed dataset: a claimed segment or place with no records, a segment more than twice the tolerance from its share, records that start after or end before the claimed period, and representative or balance wording the counts contradict; then segments past the tolerance, segments the claim leaves out, class imbalance below the floor, gaps inside the period and population shares that do not add up, with the fix for each and what would change the verdict. No segment, place or period claimed, no records per segment, an unstated population share, no records per place, no first and last records and a balance claim with no class counts leave coverage unverified. This reads one dataset against its population: the topics an agent’s knowledge covers are wever_knowledge-coverage-check and an evaluation set against the categories an agent must handle is wever_evaluation-dataset-check. Verdict dataset_coverage_skewed, dataset_coverage_partial, dataset_coverage_unverified or dataset_coverage_representative. POST /api/dataset-coverage-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-freshness-checkAgent Datasets: submit a dataset’s freshness claims with its update records: the claim with the last update it states, its documented cadence (hourly, daily, weekly, monthly, quart…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,664 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s freshness claims with its update records: the claim with the last update it states, its documented cadence (hourly, daily, weekly, monthly, quarterly, yearly or irregular), the date the data is said to run to and its wording, read for real-time, live or always current; the update log with each update’s id, date or time, partition, records and kind (full, incremental or a correction); each partition with its name, last update and newest record; the date freshness is read at; and optionally the cadence periods an update may run late (1 when left out). The signed report gives the cadence with its period, the grace allowed, the median and longest gaps the log shows and the wording found, the updates with the first and latest logged, the stated last update, the days since the latest at the reading date, the as-of date, the newest record and every gap longer than the cadence allows, every partition as current, stale, tail_short or unread with the days since its update, and counts them, and names stale data: an update overdue at the reading date, gaps in the log longer than the cadence allows, a stated last update later than the log, a partition past the cadence, an as-of date later than the newest record and real-time wording a slower cadence contradicts; then a stated last update earlier than the log, partitions whose records stop before the as-of date and updates dated after the reading date, with the fix for each and what would change the verdict. No update log, cadence, stated last update or reading date leaves freshness unverified. This reads one dataset’s freshness against its update records: stale versions named in documentation are wever_docs-freshness-check, dated knowledge entries are wever_knowledge-freshness-check and cached context that can outlive its source is wever_cache-staleness-check. Verdict dataset_freshness_stale, dataset_freshness_partial, dataset_freshness_unverified or dataset_freshness_current. POST /api/dataset-freshness-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-license-checkAgent Datasets: submit a dataset’s license and usage rights with the intended use: the license with its name, SPDX id (such as CC-BY-4.0, CC-BY-NC-4.0 or ODbL-1.0), its text, read …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,634 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s license and usage rights with the intended use: the license with its name, SPDX id (such as CC-BY-4.0, CC-BY-NC-4.0 or ODbL-1.0), its text, read for wording such as non-commercial, research only, no derivatives, no redistribution, share alike, attribution required and no model training, the stated terms for commercial use, redistribution and derived works (each allowed, prohibited, restricted or not known), whether it asks for attribution or the same terms downstream, and any other restrictions as stated; the use with its purpose and whether it is commercial, redistributes the data, makes derived works, trains models, plans attribution and releases derived work under the same terms; and each upstream source inside the dataset with its id, name, license and stated terms. The signed report gives the license as recognized with the wording found, every term the use needs as allowed, conflicts, conditional, met, unmet, unread or not_needed with whether it came from the stated term, the SPDX license or the wording, every upstream source as clear, conflicts or unread, and counts them, and names a conflict: a commercial use, redistribution, derived work or model training the license prohibits, share-alike or attribution duties the use does not plan and an upstream source whose terms rule the use out; then commercial use the license allows only under conditions and stated terms that contradict the named license or its text, with the fix for each and what would change the verdict. An unidentified license and a term the use needs that neither the license nor a source settles leave the rights unverified. This reads one dataset’s terms against one use: the license family a text matches is wever_license-evidence-check and the purposes a consent grant covers are wever_consent-purpose-check. Verdict dataset_license_conflicts, dataset_license_partial, dataset_license_unverified or dataset_license_clear. POST /api/dataset-license-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-lineage-checkAgent Datasets: submit a dataset’s lineage claims with its source records: optionally the claim with the sources it says the dataset draws from, the release license and its wording…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,689 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s lineage claims with its source records: optionally the claim with the sources it says the dataset draws from, the release license and its wording, read for fully traceable, complete lineage or end-to-end provenance; each source with its id, name, license by name or SPDX id, where it was obtained, when, its stated commercial terms and whether it asks for attribution; each transformation step in order with its id, name, the source or earlier step ids it reads, whether its code or run log is recorded and whether upstream terms travel with its output; and the output with the step or source the released dataset comes from and its license. The signed report gives every source with its license, origin, whether it feeds the release, whether the claim names it and its issues, every step with its inputs, any input nothing provides and whether it is recorded and carries terms, the chain with the release, whether it is reached, the sources reaching it, the release license and its terms and the wording found, and counts them, and names a broken lineage: a source the claim names with no record, a source feeding the release that the claim leaves out, a step reading input no source or earlier step provides, a release no step reaches, a source whose terms rule out commercial use behind a release that allows it, an attribution duty a step drops and traceability wording the chain does not hold; then steps with no recorded code or run log and sources no step reads, with the fix for each and what would change the verdict. No steps, an unstated source license or origin, an unstated record of a step, no claimed sources and no release license leave the lineage unverifiable. This traces one dataset from its sources: the sources behind a knowledge base are wever_knowledge-source-check and where stored memories came from is wever_memory-provenance-check. Verdict dataset_lineage_broken, dataset_lineage_partial, dataset_lineage_unverifiable or dataset_lineage_complete. POST /api/dataset-lineage-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-privacy-checkAgent Datasets: submit a dataset’s privacy claims with its contents and handling records: the claim with whether the dataset is said to be anonymized and whether it holds personal …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,070 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s privacy claims with its contents and handling records: the claim with whether the dataset is said to be anonymized and whether it holds personal data, and its wording, read for anonymized, de-identified, no personal data or non-identifiable; each field with its name, kind (a direct identifier, a quasi-identifier, sensitive or other) and treatment (removed, hashed, tokenized, encrypted, masked, generalized or none); a sample of records, each with its id and its values as written; the handling record with the legal basis (consent, contract, legal obligation, vital interests, public task, legitimate interests or none stated), whether consent is recorded, the days data is kept, the deletion process and who can reach the data (public, restricted or internal); and optionally the smallest group a quasi-identifier combination may form (5 when left out). The signed report gives every field with its kind, treatment, the identifier kinds found in its values and its status as exposed, pseudonymized, sensitive_open or treated, the identifiers found by kind, field and record with the values never repeated, the re-identification reading with the groups the quasi-identifiers form in the sample, the smallest group and the groups below the threshold, the handling with the basis, consent, retention, deletion and access, and counts them, and names exposed data: an untreated direct identifier, an identifier-shaped value where none should remain, quasi-identifier groups below the threshold under an anonymity claim, personal data the claim denies, no legal basis, unrecorded consent and untreated sensitive fields in public data; then direct identifiers only hashed, tokenized or encrypted under an anonymity claim, no deletion process and small groups with no anonymity claim, with the fix for each and what would change the verdict. No fields or sample, an unstated kind or treatment, legal basis, retention or access, and declared quasi-identifiers with no sample leave privacy unverified. This reads one dataset’s contents: privacy claims against a policy document are wever_privacy-evidence-check, a retention and deletion policy is wever_data-retention-check and the purposes a consent grant covers are wever_consent-purpose-check. Verdict dataset_privacy_exposed, dataset_privacy_partial, dataset_privacy_unverified or dataset_privacy_protected. POST /api/dataset-privacy-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-quality-checkAgent Datasets: submit a dataset’s stated quality claims with its own sample and profile: optionally the dataset with its id, name, version and the rows it states; the claims of th…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,838 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s stated quality claims with its own sample and profile: optionally the dataset with its id, name, version and the rows it states; the claims of the highest share of empty values for any field, the highest share of duplicate records, the highest share of invalid values and the label accuracy, with the quality wording, read for absolutes such as no missing values, fully deduplicated or 100% accurate; each field with its name, type (string, integer, number, boolean, date or category), whether it is required, its allowed values and its range; the labels with the label field, the label set and how many labels a review checked and agreed with; the profile with its row count, duplicate rows and each field’s empty and invalid counts; and a sample of records, each with its id, its values as written (a value left out reads as an empty cell) and its label. The signed report gives every field with its empty and invalid shares in the sample and the profile, the measures with the highest empty, duplicate and invalid shares beside the shares claimed, the labels with those outside the set, identical records labeled differently and the accuracy the review measured, and counts them, and names a deficient dataset: a field empty more often than claimed, a required field left empty, duplicates or invalid values above the claim, label accuracy below the claim, identical records labeled differently and absolute wording the data contradicts; then labels outside the set, fields the declaration leaves out and a row count that differs from the profile, with the fix for each and what would change the verdict. No sample or profile, no claimed figure, an untyped field and an accuracy claim no review backs leave the quality unverified. This reads one dataset’s claims against its own data: one JSON sample checked field by field is wever_data-evidence-check, an evaluation set against what an agent must handle is wever_evaluation-dataset-check and a claim against offered evidence is wever_claim-evidence-check. Verdict dataset_quality_deficient, dataset_quality_partial, dataset_quality_unverified or dataset_quality_sound. POST /api/dataset-quality-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-schema-drift-checkAgent Datasets: submit a dataset’s current schema with its documented or prior schema: the documented schema with its version, where it comes from (a data dictionary, a schema file…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,705 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s current schema with its documented or prior schema: the documented schema with its version, where it comes from (a data dictionary, a schema file, a prior release, a data contract, a readme or other) and each field with its name, type, whether it may be null, whether it is required and its description; the current schema with its version and the same fields; the renames the release records, each from a documented field to a current one; the readers with each one’s name and the fields it reads; and the change with whether the release was declared none, compatible or breaking, its version bump (none, patch, minor or major) and its release notes, read for wording such as no breaking changes or backward compatible. The signed report gives every field as unchanged, added, removed, renamed, retyped, type_widened, nullable_now or required_now and as breaking or compatible for a reader, the changes with both versions, the declaration, the bump and whether it fits the drift, every reader with the fields a breaking change touches, and counts them, and names a broken schema: a removal, rename, narrowing type change or newly nullable field shipped without a breaking declaration and a major version or under notes that deny it, a reader that reads a broken field and a recorded rename with no field behind it; then breaking changes declared with a major version that readers must still move to, and compatible changes shipped under no declaration or no version bump, with the fix for each and what would change the verdict. An unstated declaration, bump, field type or reader list leaves the drift unchecked. This reads one dataset schema against its documented one: two versions of one text compared clause by clause are wever_contract-diff-check, behavior against a written spec is wever_spec-drift-check and two evaluation runs compared are wever_evaluation-drift-check. Verdict dataset_schema_broken, dataset_schema_partial, dataset_schema_unchecked or dataset_schema_stable. POST /api/dataset-schema-drift-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_dataset-synthetic-disclosure-checkAgent Datasets: submit a dataset’s synthetic-data disclosure with its provenance records: the disclosure with the synthetic share it states, the generation method it names (such as…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,616 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
Description, as the server wrote it Agent Datasets: submit a dataset’s synthetic-data disclosure with its provenance records: the disclosure with the synthetic share it states, the generation method it names (such as an LLM, a GAN or rule-based simulation), the generator, the field that marks a record synthetic and the disclosure as published, read for wording such as real-world data only or no synthetic records; the provenance with the total and synthetic record counts and each generation run with its id, method, generator, records and date; a sample of records, each with its id, whether provenance records it as synthetic and whether it carries the marker; and optionally the tolerance in points (2 when left out). The signed report gives the disclosure with its share, method, generator, marker and wording, the provenance with the share its counts or runs give and the methods the runs record, the marking with the records read, synthetic records left unmarked, real records marked synthetic and whether synthetic records can be told apart, and counts them, and names an undisclosed share: synthetic records with no stated share, a stated share below the records, wording that presents the data as wholly real, synthetic records with no marker field and synthetic records the marker misses; then an unnamed generation method, a method other than the runs record, a stated share above the records and real records marked synthetic, with the fix for each and what would change the verdict. No provenance, an unstated total or synthetic count and a sample that states no provenance or marker leave the disclosure unverifiable. This reads one dataset’s disclosure against its provenance: AI involvement in content or a product is wever_ai-disclosure-check and a claim against offered evidence is wever_claim-evidence-check. Verdict dataset_synthetic_undisclosed, dataset_synthetic_partial, dataset_synthetic_unverifiable or dataset_synthetic_disclosed. POST /api/dataset-synthetic-disclosure-check. Reads the dataset records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Datasets reads a dataset’s own records and checks what is claimed about it: its quality claims against its own sample and profile, its current schema against the documented or prior one, its license against the intended use, its synthetic-data disclosure against its provenance, its freshness against its update log, its lineage against its source records, its privacy claims against its contents and handling, and its coverage against the population it is said to represent. Beside it, wever_data-evidence-check inspects one JSON sample field by field and wever_evaluation-dataset-check reads an evaluation set against what an agent must handle, while a dataset check reads a claim made about a dataset against the dataset’s own records; wever_license-evidence-check names the license family a text matches, while the license check reads the terms against one intended use; and wever_privacy-evidence-check reads privacy claims against a policy document, while the privacy check reads the dataset’s contents and handling records. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_deadline-checkWever Checks: submit document text (up to 256 KB) and get a signed report of every date and obligation in it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Wever Checks: submit document text (up to 256 KB) and get a signed report of every date and obligation in it. Each obligation is quoted with its offset, party, action, category and due date, relative period or unparsed due date; a timeline shows what is due when. Dates that could not be parsed (no year, no day, two-digit year, ambiguous day and month, or not a calendar day) are quoted with the reason. POST /api/deadline-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/checks-contract. Human page: https://weverlabs.com/checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegate-vetting-checkAgent Delegation: submit a delegator’s selection flow: the delegate chosen, the task and when it was handed over, how the delegate’s identity was checked (verified, taken on its ow…3 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...ainst what (a registry record, a signed credential, an attester, a prior working relations..."
- Info
The input schema at properties.identity.properties.method.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.identity.properties.method.description: "...ed against: a registry record, a signed credential, an attester, a prior working relations..."
- Info
The description is 2,020 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
Description, as the server wrote it Agent Delegation: submit a delegator’s selection flow: the delegate chosen, the task and when it was handed over, how the delegate’s identity was checked (verified, taken on its own word, or not checked) and against what (a registry record, a signed credential, an attester, a prior working relationship or other), whether a track record was looked at, where it came from (a public record, an attester, the delegator’s own history or the delegate’s own claims), whether it covers comparable work and was public before the delegate’s first engagement, and whether fit for this task was checked, with what the task needs and what the delegate showed, each with when it was checked. The signed report says whether the delegate was checked before the handoff on identity, track record on comparable work and fit for this task specifically, and names an unchecked or self-declared identity, no track record looked at and every check done after the handoff, then a self-reported or non-comparable record, a record assembled after first use, fit not checked and every need not shown met, with the fix for each. This reads the delegator’s diligence before the handoff: whether a purchase receipt’s buyer identity matches its delegation is wever_delegated-identity-check, one identity record’s own consistency is wever_agent-identity-check and whether a handoff route checks its receiver against the registry is wever_receiver-identity-check. Verdict stranger_handoff, vetting_gaps, vetting_unverified or delegate_vetted. POST /api/delegate-vetting-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegated-identity-checkAgent Purchases: submit one purchase receipt (the buyer identity it names and its kind, the party it was paid for, the acting agent, merchant, category, amount, time and the creden…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...erchant, category, amount, time and the credential paid with) and the delegation behind it..."
- Info
The input schema at properties.receipt.properties.credential.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.receipt.properties.credential.description: "The credential the payment was made with."
Description, as the server wrote it Agent Purchases: submit one purchase receipt (the buyer identity it names and its kind, the party it was paid for, the acting agent, merchant, category, amount, time and the credential paid with) and the delegation behind it (who delegated, to which agent, the merchants, categories and per-payment amount it allows, when it expires, how it is revoked and any revocation time). The signed report says whether the buyer identity on the receipt is the delegating business or person and the acting agent the delegated one, names an agent paying as itself, a receipt identity the delegation does not cover and a purchase after revocation, then every purchase outside the delegation’s merchants, categories or amount, an unscoped or expired delegation, a delegation that cannot be revoked and an agent paying with the business’s own key. Issuing a mandate is wever_delegated-authority, a delegation document’s own terms are wever_delegation-check and an agent identity record’s consistency is wever_agent-identity-check. Verdict pays_as_itself, not_delegated, delegation_gaps, identity_unverified or properly_delegated. POST /api/delegated-identity-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-chain-checkAgent Delegation: submit a delegation chain as supplied: the principal at the top, the work it serves, whether the whole chain is kept as one record, and each hop in order with who…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Delegation: submit a delegation chain as supplied: the principal at the top, the work it serves, whether the whole chain is kept as one record, and each hop in order with who delegates to whom, the tasks, spending ceiling, tools, data and end passed on, and whether the principal can see that hop. The signed report lays the hops end to end, says hop by hop whether the scope narrowed, held or grew against the hop above, whether the scope narrowed overall and whether the chain’s full length is visible to the principal, and names every hop that passes on tasks, tools or data the hop above lacks, a higher ceiling or a later end, and every hop that drops a bound the hop above states, then every hop out of the principal’s sight, every hop that does not link to the one above, a first hop that does not start from the principal and a chain kept in no one record, with the fix for each. This reads authority as it passes down a chain of delegations: who held one task at each step is wever_chain-of-custody-check, approver bands are wever_approval-chain-check and one handoff’s grant against what its task needs is wever_handoff-scope-check. A summary sample is wever_delegation-chain-sample on https://weverlabs.com/mcp-free. Verdict authority_grew, chain_gaps, chain_unverified or authority_narrowed. POST /api/delegation-chain-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-checkAgent preflight: check a delegation or mandate document before you act under it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight: check a delegation or mandate document before you act under it. Input text, fields (grantee, scope as text or actions with their own cap, expiry and approval, caps, expiry, revocation, approvals) or both. The signed report marks the grantee, scope of actions, caps, expiry, revocation terms and approval requirements present, missing or ambiguous, with quotes, and names every delegated action with no cap and every term with no expiry. POST /api/delegation-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-expiry-checkAgent Delegation: submit a delegation setup’s expiry: what ends it (a stated date, a length from issue, the task being done, or nothing) with when it was issued, the end date or th…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,053 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
Description, as the server wrote it Agent Delegation: submit a delegation setup’s expiry: what ends it (a stated date, a length from issue, the task being done, or nothing) with when it was issued, the end date or the length, when the expiry is checked (before each use, at the start of each task, only at setup, or never), what a lapsed delegation does when used anyway (refused and the principal told, refused, allowed and logged, or allowed), how it is renewed (by the principal on purpose, by itself, or not at all) and whether the principal is warned before it lapses, plus optionally the uses seen with when each happened and whether it went ahead, and the date they are read at. The signed report says whether an expiry is set, whether it is checked before each use or only at setup and what a lapsed delegation does when used anyway, with its end and days left, and names no expiry, a length of forever, an expiry checked only at setup or never, a lapsed use allowed, every use that went ahead after the end and a delegation that renews itself, then an expiry checked only per task, a lapsed use refused without telling the principal, no warning before it lapses and an end with no date or length, with the fix for each. This reads the end set in advance and its enforcement: pulling a delegation back early is wever_delegation-revocation-check, whether a delegation document states an expiry at all is wever_delegation-check and a handoff grant that outlives its task is wever_handoff-scope-check. Verdict never_expires, expiry_gaps, expiry_unverified or expiry_enforced. POST /api/delegation-expiry-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-record-checkAgent Delegation: submit a system’s delegation records (each with its id, when the delegation was made, who delegated, who received it, what was delegated, the scope given, what ca…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Delegation: submit a system’s delegation records (each with its id, when the delegation was made, who delegated, who received it, what was delegated, the scope given, what came back, how it was settled: paid, accepted, sent back, disputed, cancelled or still open, and when the record was written) and optionally the delegations seen in other logs, such as payments, messages or a tracker. The signed report counts every delegation that keeps the essentials and names every record missing who delegated, who received it or what was delegated (the core a delegation cannot be audited without), every record missing the scope given, what came back or how it was settled, every record written more than a day after the delegation, and every delegation seen elsewhere with no record behind it, with the fix for each. This reads the record of delegations specifically: the record of handoff events is wever_handoff-record-check, the record of interruptions and recoveries is wever_recovery-record-check and the buyer’s record of paid calls is wever_purchase-record-check. Verdict delegations_unrecorded, record_gaps, records_unverified or delegations_recorded. POST /api/delegation-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-reporting-checkAgent Delegation: submit a delegation setup’s reporting duty: whether the delegate owes the delegator reports, how often (such as every 4 hours or daily), in what format (a structu…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,217 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
Description, as the server wrote it Agent Delegation: submit a delegation setup’s reporting duty: whether the delegate owes the delegator reports, how often (such as every 4 hours or daily), in what format (a structured record, a fixed template, free text or none), what each report carries and when failures must be reported (as they happen, in the next scheduled report, at the end of the task, or not at all), how a missed report is noticed (whether it is watched, how soon, and whether the principal or an operator is alerted, the delegation paused, the delegate reminded or nothing happens), plus optionally the reports seen with when each arrived, whether it was scheduled, a failure report or the final one, when a reported failure happened, and the date they are read at. The signed report says whether the delegate owes reports on a stated cadence and format, whether failures must be reported as they happen and whether a delegate that goes quiet is noticed, and names no duty to report, failures held to the end or never reported and silence nobody notices or that changes nothing, then no cadence, no or free-text format, failures held for the next report, a reminder as the only response, slow notice, every gap between reports past the cadence, a delegate quiet now and every failure reported more than an hour late, with the fix for each. This reads the delegate’s standing reporting duty inside a delegation: notice to the principal when a task fails for good is wever_failure-notice-check, where a refused, failed or silent handed-off task goes is wever_handoff-failure-check and task state reporting during recovery is wever_partial-state-check. Verdict silence_passes, reporting_gaps, reporting_unverified or reports_owed. POST /api/delegation-reporting-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-result-checkAgent Delegation: submit a delegation’s return flow: whether the original brief is kept with what counts as done, what returned work is checked against (the brief, the delegate’s o…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,085 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
Description, as the server wrote it Agent Delegation: submit a delegation’s return flow: whether the original brief is kept with what counts as done, what returned work is checked against (the brief, the delegate’s own report, or nothing) and by whom (the delegator, an independent checker, the delegate itself, or no one), whether the check comes before the work is accepted, paid for and passed on, and what happens to work that fails (sent back with the reasons, escalated, redone by the delegator, dropped without a word, or accepted anyway), plus optionally the returns seen with whether each was checked and passed, the criteria it missed and whether it was accepted, paid for or passed on. The signed report says whether the delegator verifies returned work against the original brief before accepting it, paying for it or passing it on, and what happens to work that fails, and names no check, a check by the delegate or against its own report, no brief kept, every use before the check, failed work accepted, every return used unchecked and every failed return used, then a brief with no criteria, failed work redone by the delegator or dropped without a word and every return passed with criteria unmet, with the fix for each. This reads acceptance of a delegate’s work inside a delegation: one paid call’s result against its order is wever_result-validation-check, a job’s work against acceptance criteria is wever_deliverable-check and who holds a task between send and acceptance is wever_handoff-acceptance-check. Verdict accepted_unchecked, result_gaps, result_unverified or checked_against_brief. POST /api/delegation-result-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-revocation-checkAgent Delegation: submit a delegation setup: who can revoke it (the principal, the principal or an operator, only an operator, only the delegate, or no one) and how the principal d…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,180 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
Description, as the server wrote it Agent Delegation: submit a delegation setup: who can revoke it (the principal, the principal or an operator, only an operator, only the delegate, or no one) and how the principal does (on their own, by asking, or not at all), whether it can be revoked while a task is under way, when a revocation takes hold (at once, at the delegate’s next call, at its next task, at the next sync or by hand) and how long it takes against a stated target, what happens to work in flight when it lands (halted, rolled back, allowed to finish the step, or left running on the old authority), what happens to authority the delegate passed on and whether the principal is told, plus optionally the revocations seen with when each was made and took hold and whether work went on after. The signed report says whether the principal can revoke the delegation mid-task, how fast the revocation takes effect and what happens to work already in flight, and names no way to revoke, revocation only by an operator or the delegate, revocation only between tasks, work left running and every revocation work went on after, then revocation on request, a revocation deferred to the next task, a sync or a hand step, a revocation slower than the target, authority passed on that survives, no confirmation and every revocation that took hold late, with the fix for each. This reads pulling a delegation back early: the end set in advance is wever_delegation-expiry-check, halting a whole agent system is wever_kill-switch-check and whether a delegation document states revocation terms is wever_delegation-check. Verdict cannot_pull_back, revocation_gaps, revocation_unverified or revocable_mid_task. POST /api/delegation-revocation-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_delegation-scope-checkAgent Delegation: submit one delegation as supplied: who delegates to whom, where its bounds are written (a delegation record both sides read, a message to the delegate, only code …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,120 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
Description, as the server wrote it Agent Delegation: submit one delegation as supplied: who delegates to whom, where its bounds are written (a delegation record both sides read, a message to the delegate, only code or configuration, or nowhere), when they were written and when work started, the tasks the delegate may do and whether the list ends in open wording, whether it may spend with the ceiling and what it counts over (the whole delegation, each task, each payment, a day, a week or a month), the tools and data it may use, and what ends it (the task being done, a date, a length of time, only a revocation, or nothing). The signed report says bound by bound whether the scope is stated before work starts, which tasks, what spend, which tools and data and for how long, and names bounds written nowhere, bounds written after work started, no named tasks, open task wording, spending with no ceiling and no end, then bounds set only in code, unstated spending, a ceiling with no period, unnamed or open tools and data, an end only on revocation and an end with no date or length, with the fix for each. This audits the written bounds of one delegation from the delegator’s side: whether a delegation or mandate document marks its grantee, caps, expiry, revocation and approvals present is wever_delegation-check, issuing a signed mandate is wever_delegated-authority and what each receiver gets against what its task needs in an agent-to-agent handoff is wever_handoff-scope-check. A one-record restatement is wever_delegation-snapshot on https://weverlabs.com/mcp-free. Verdict blank_check, scope_gaps, scope_unverified or scope_bounded. POST /api/delegation-scope-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_deliverable-checkDiligenceOps: submit a job’s acceptance criteria and the delivered work’s text, metadata or both, with optional supporting evidence.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a job’s acceptance criteria and the delivered work’s text, metadata or both, with optional supporting evidence. The signed checklist marks each criterion addressed, missing or not met, measuring counts, sizes, durations and file formats where the delivery states them, and lists claims in the deliverable (figures, certainty, superlatives) that the supplied material does not support. POST /api/deliverable-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_deterministic-core-checkAgent Control: submit a system design that splits work between a model and code: each decision or computation with its kind (arithmetic, threshold, eligibility, limit, rounding, co…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Control: submit a system design that splits work between a model and code: each decision or computation with its kind (arithmetic, threshold, eligibility, limit, rounding, conversion, date, formatting, extraction, classification and others), who makes it (model, code, or model checked by code), the check applied and what the output is used for, plus the deterministic components available. The signed report names every decision the model makes that code should compute, every model output used for money, a customer or a record with no deterministic check, and the smallest reassignment for each. Verdict model_computes, unchecked_output or deterministic_core. POST /api/deterministic-core-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_digest-createStanding jobs: register a digest for the paying wallet with a cadence (hourly, daily, weekly or interval), an optional start_at, a callback_url and an expires_at up to 30 days ahea…Reads text others writeCan send data outNo findings
Why it is in this group
- The name "wever_digest-create" says "create": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Standing jobs: register a digest for the paying wallet with a cadence (hourly, daily, weekly or interval), an optional start_at, a callback_url and an expires_at up to 30 days ahead. Each window collects the wallet’s Wever Watch fires and standing-job results recorded while it was open (sources: watch_fires, job_results or both) and posts one signed digest report to callback_url. Windows with nothing to report are posted only with include_empty: true. One call covers every window until the digest ends. POST /api/digest-create. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_diligenceops-paid-runPaid URL: 1.00 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- The name "wever_diligenceops-paid-run" says "run": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...diligenceops-paid-run. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 1.00 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Free computation: /api/mcp or /api/service-resolver. Read the backend 402 body for x402 terms. Compare submitted evidence labels with the required list and return matched and missing items. Operating boundary: Supply required and available evidence labels. Read matched and missing labels. An agent uses the result to request the missing items; a human reviews the list and checks the documents. Direct runs cost 1 USDC after the shared free allowance; the Labs MCP and resolver computation paths are free. POST /api/diligenceops-paid-run. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_directory-listing-checkAgent directory round 2: check a directory listing draft before you submit it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent directory round 2: check a directory listing draft before you submit it. Input the fields wever_directory-submit accepts: name, website, optional mcp_url, description, category (the directory layer), access (free, freemium or paid) and an optional email, which is checked and never repeated. The signed report evaluates the draft against the standards the Wever directory publishes under: the form field rules, a public HTTPS address for the website and endpoint, a description that says what the service does, the pricing stated, an MCP endpoint whenever the listing claims one, the content screen and the duplicate rule against live listings, with the specific fixes that would make it pass. It submits and publishes nothing. POST /api/directory-listing-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-records-contract. Human page: https://weverlabs.com/directory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_directory-submitAgent directory: list a server or tool in the free Wever agent directory from your agent.No findings
Why it is in this group
- The name "wever_directory-submit" says "submit": a word for a change.
Description, as the server wrote it Agent directory: list a server or tool in the free Wever agent directory from your agent. Input the directory form fields: name, website, optional mcp_url, description, category (the directory layer) and access (free, freemium or paid), plus a contact email kept private. The web form validation runs first; any field that needs a change comes back by name with its rule, nothing is queued and the call is not billed. A complete listing is visible immediately in the page, feed and search, with review pending. The hourly run checks it afterward and reports concerns without automatically removing it. Edits require the matching original contact email. Returns a signed record with the submission_id and status published. POST /api/directory-intake. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-records-contract. Human page: https://weverlabs.com/directory/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_discovery-coverage-checkAgent Governance: submit a seller’s agent-discovery setup: the offer with its price, terms and last change, and every surface where buying agents find it (registries, directories, …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Governance: submit a seller’s agent-discovery setup: the offer with its price, terms and last change, and every surface where buying agents find it (registries, directories, agent cards, feeds, marketplaces and MCP servers), each with whether it lists the offer, the price and terms it shows, when it last took the offer, how changes reach it and which agents find the seller only there. The signed report names every surface where the offer is missing, stale or contradicts the offer or another surface on price or terms, and every single surface whose failure makes the seller invisible to buying agents. One listing’s quality is wever_listing-check and how a router chooses a seller is wever_router-disclosure-check. Verdict contradictory, coverage_gaps, coverage_unverified or consistent. POST /api/discovery-coverage-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_docs-freshness-checkDiligenceOps: submit documentation (text, or a public HTTPS url) and the current_version.Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it DiligenceOps: submit documentation (text, or a public HTTPS url) and the current_version. The signed report counts and briefly quotes mentions of older and newer version numbers, including /v1/ style paths, and the endpoints, parameters and functions the text itself marks deprecated, with how often each is still named elsewhere in the text. POST /api/docs-freshness-check. Reads the supplied text, or one public HTTPS page read once with GET (no redirects, public DNS only, 512 KB): no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_dpa-evidence-checkDiligenceOps: submit a counterparty’s data processing agreement or its key clauses, as text, fields (roles, purpose_limits, subprocessors, breach_notification, transfers, deletion_…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a counterparty’s data processing agreement or its key clauses, as text, fields (roles, purpose_limits, subprocessors, breach_notification, transfers, deletion_return) or both. The signed report marks the roles of the parties, purpose limits, sub-processor terms, the breach notification window, the transfer mechanism and deletion or return on termination present, missing, ambiguous or one-sided, with quotes: processing for the processor’s own purposes, sub-processors added at its discretion with no right to object, a breach window longer than 72 hours, transfers anywhere with no mechanism and copies kept after termination. POST /api/dpa-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_duplicate-effect-checkAgent Recovery: submit a resume and retry flow: each effect a resumed or retried task can repeat (a message, an order, a record, a booking or something else) with how a repeat is s…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Recovery: submit a resume and retry flow: each effect a resumed or retried task can repeat (a message, an order, a record, a booking or something else) with how a repeat is stopped or caught (an idempotency key the receiver honors, a check for the effect before repeating it, a cleanup that removes repeats afterward, or nothing), what one idempotency key covers (the effect, the task or each attempt), whether the flow looks for the effect after a resume or retry and whether the receiver drops repeats on its own, plus optionally the recovered runs with how many times each effect happened. The signed report says effect by effect whether duplicated effects are checked for after recovery, and names every effect nothing guards, every idempotency key issued per attempt and every effect seen happening more often than expected (messages sent twice, orders placed twice, records created twice, bookings doubled), then every repeat only cleaned up afterward and every effect not looked for after recovery, with the fix for each. This reads every effect a task can duplicate other than payment, which it lists without a judgment: a payment charged twice is wever_double-charge-check, one timed-out checkout against the merchant’s order records is wever_duplicate-order-check and whether one payment or refund flow’s retries are safe is wever_retry-safety-check. Verdict effects_can_repeat, duplicate_gaps, duplicate_unverified or duplicates_checked. POST /api/duplicate-effect-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_duplicate-order-checkJob Records: submit one timed-out or ambiguous checkout (merchant, item, amount, timestamp and the order reference where known) with the merchant order or receipt records you hold …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Job Records: submit one timed-out or ambiguous checkout (merchant, item, amount, timestamp and the order reference where known) with the merchant order or receipt records you hold and, where you have it, when those records were exported. The signed report compares each record field by field and returns matching_order_found when a completed order already covers the checkout (a second order would duplicate it), no_matching_order when a fresh order is consistent with the records, or indeterminate when the records cannot settle it, with the matched fields itemized and the open points named. Records comparison only; the merchant is never contacted and no order is placed. POST /api/duplicate-order-check. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_endpoint-honesty-checkAgent records round 2: check that an x402 endpoint charges what it advertises.Reads text others writeCan send data out1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...nly, 5 seconds, a response size cap, no credentials) and never pays it. The signed report s..."
Description, as the server wrote it Agent records round 2: check that an x402 endpoint charges what it advertises. Input url (HTTPS), price in whole units, asset (a token address, or USDC on Base and Base Sepolia), network (CAIP-2 or an x402 version 1 name), optional pay_to, optional asset_decimals and optional label. Wever reads the endpoint’s live 402 challenge with one unpaid GET through the endpoint reliability guards (public HTTPS addresses only, 5 seconds, a response size cap, no credentials) and never pays it. The signed report says whether the live amount, asset, network and recipient match the advertised terms and names each mismatch with both values. Availability and latency stay with wever_endpoint-reliability-check. POST /api/endpoint-honesty-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-records-contract. Human page: https://weverlabs.com/reliability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_endpoint-reliability-checkEndpoint reliability: check one MCP server endpoint or x402 HTTP endpoint live.Reads text others writeCan send data out1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...as anonymous access, payment required, credential required or unreachable; and the checke..."
Description, as the server wrote it Endpoint reliability: check one MCP server endpoint or x402 HTTP endpoint live. Input url (HTTPS), optional label and optional kind (auto, mcp or x402). The signed report gives a reliability score from 0 to 100 built from reachability, TLS, protocol, latency and consistency; the HTTPS status code; the TLS certificate result; for MCP endpoints, the initialize and tools/list handshake the Wever directory runs for listings, with the tool count and whether the list is the same on a second call; for x402 endpoints, whether the 402 response carries a well-formed payment challenge; the median latency of three requests; the auth behavior as anonymous access, payment required, credential required or unreachable; and the checked-at time. It reads public HTTPS addresses only, with a 5 second limit per request and a response size cap, and sends no credentials or payment. POST /api/endpoint-reliability-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-records-contract. Human page: https://weverlabs.com/reliability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_energy-bill-checkAgent Energy: submit a utility bill with its tariff and usage record: the tariff with its name, each rate with its id, name, component, unit (per kWh, per kW, per day or per month)…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,908 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
Description, as the server wrote it Agent Energy: submit a utility bill with its tariff and usage record: the tariff with its name, each rate with its id, name, component, unit (per kWh, per kW, per day or per month), time-of-use period, rate and for a tiered kWh rate the usage the tier starts after and ends at, and each tax with its id, name, rate and what it is charged on (all charges, energy or delivery); optionally the usage record with its period, total kWh, on-peak, mid-peak and off-peak kWh, peak kW, opening and closing meter reads and whether the read was actual or estimated; the bill with its id, period, each line with its id, description as printed, the tariff rate it charges, quantity, rate printed and amount, each tax line with its tariff tax, name and amount, the total and currency; and optionally the tolerance (0.01 when left out). The signed report gives the usage, every line beside the tariff rate times the recorded usage as matches, overcharged, undercharged, misstated, not_in_tariff, duplicate, unmatched or unread with its expected quantity, rate and amount, every tax line beside its recomputed amount, the totals billed, summed and recomputed with the difference, and counts them, and names errors: a line the tariff does not list, a repeated charge, a line or tax above its recomputed amount, a tax the tariff does not apply, a total above its lines and usage that disagrees with the meter reads; then an undercharge, a misprinted rate or quantity, a tariff charge or tax the bill leaves off, a total below its lines, time-of-use parts that do not add up, an estimated read and a period that differs from the usage record, with the fix for each and what would change the verdict. A missing usage record, usage field, rate reference or total leaves the bill unverified. This recomputes one energy bill: an invoice against the job it belongs to is wever_invoice-check, one invoice summarized is wever_invoice-brief, a fee schedule against a general invoice is wever_fee-evidence-check and the plan itself against the plan quoted is wever_energy-tariff-check. Verdict energy_bill_errors, energy_bill_partial, energy_bill_unverified or energy_bill_clean. POST /api/energy-bill-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_energy-carbon-claim-checkAgent Energy: submit a carbon credit or renewable certificate claim and the instruments behind it: the claim as made, read for absolute wording such as carbon neutral, net zero or …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,627 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
Description, as the server wrote it Agent Energy: submit a carbon credit or renewable certificate claim and the instruments behind it: the claim as made, read for absolute wording such as carbon neutral, net zero or 100% renewable, with the claimant, what backs it (carbon offsets or renewable energy certificates), the quantity and unit it covers, its period and when it was made; optionally the footprint it covers with its quantity, unit and basis; each instrument with its id, registry, serial, project, kind, vintage, quantity, unit, status (retired, active, cancelled, transferred or not known), when and for whom it was retired and any other party that claims it (an empty list when none); and optionally the most years a vintage may sit from the claim period (1 for certificates and 5 for offsets when left out). The signed report gives the claim with its wording, every instrument with its issues and whether it backs the claim, the coverage with the quantity claimed, the quantity retired behind it and the footprint, and counts them, and names an unbacked claim: no instrument, one not retired, retired for someone else, claimed by another party or listed twice, a retired total below the claimed quantity and absolute wording that reaches past what was retired; then an instrument in another unit than the claim, a vintage far from the claim period and a retirement after the claim was made, with the fix for each and what would change the verdict. An untraced instrument and an unstated status, beneficiary, quantity, footprint or period leave the claim unverifiable. This reads one claim against its retirement records: sustainability claims against offered evidence are wever_environmental-evidence-check and a product claim against its evidence is wever_claim-evidence-check. Verdict energy_carbon_unbacked, energy_carbon_partial, energy_carbon_unverifiable or energy_carbon_backed. POST /api/energy-carbon-claim-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_energy-contract-checkAgent Energy: submit an energy supply contract and what was agreed before signing: the agreed terms with the supplier, term in months, rate with its unit, rate type (fixed, variabl…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,468 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
Description, as the server wrote it Agent Energy: submit an energy supply contract and what was agreed before signing: the agreed terms with the supplier, term in months, rate with its unit, rate type (fixed, variable or indexed), exit fee and its basis (none, flat or per remaining month), whether it renews by itself and onto which rate type, notice days, minimum usage (0 for none), monthly fee, start date and where the agreed terms come from; and the contract with the same terms and its text as written, read for wording about a variable rate, automatic renewal, exit fees and minimum usage. The signed report gives every term on both sides as aligned, conflicts, differs, not_agreed, not_in_contract, unit_differs or unread, with the exit fee counted as what leaving one month in would cost, the contract wording found, and counts them, and names a conflict: another supplier, term or rate type, a higher rate, exit fee, minimum usage or monthly fee, an automatic renewal no one agreed and contract wording that contradicts its own stated terms; then a lower rate or fee, another renewal rate type, a longer notice, a dropped renewal and another start date, with the fix for each and what would change the verdict. A term only one side states, units that differ and a renewal with no stated rate leave the contract unchecked. This reads one supply contract against the agreement: two versions of one text compared clause by clause are wever_contract-diff-check, one contract summarized is wever_contract-brief, a renewal notice is wever_renewal-notice-check and two versions of terms in force are wever_terms-change-check. Verdict energy_contract_conflicts, energy_contract_partial, energy_contract_unchecked or energy_contract_aligned. POST /api/energy-contract-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_energy-efficiency-claim-checkAgent Energy: submit an equipment efficiency or savings claim with the rating and the baseline: the equipment with its kind (a heat pump, an air conditioner, a furnace, a boiler, a…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,734 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
Description, as the server wrote it Agent Energy: submit an equipment efficiency or savings claim with the rating and the baseline: the equipment with its kind (a heat pump, an air conditioner, a furnace, a boiler, a water heater, lighting, a motor, a chiller or other), model, rating metric (such as SEER2, HSPF2, COP, AFUE, UEF, lumens per watt or kW per ton), rated value, where the rating comes from (a certified directory, the label, a spec sheet, the seller or not known) and its certificate; the claim with its metric and value, the savings share, the yearly kWh and money saved and its wording, read for up to and promised savings; optionally the baseline with its metric, efficiency, yearly usage, the rate the savings are valued at and what it is; optionally the rebate with its program, amount, minimum rating and metric, and each condition with whether it is met; and optionally the tolerance (5 percent when left out). The signed report gives the rating with the claimed and baseline values, the savings recomputed as a share, kWh and money with each claim as within, above_ratings or unread, the rebate with whether the rating meets its minimum and each condition, and counts them, and names an inflated claim: an efficiency above the rating, a savings share, kWh or money figure above what the rating and baseline give and a rebate the equipment does not qualify for; then a rating from the seller or an unknown source, a claim in another metric, up-to wording and promised savings, with the fix for each and what would change the verdict. A missing rating, source, baseline, usage, rate or rebate condition leaves the claim unverifiable. This reads one equipment claim against its rating: a claim against offered evidence is wever_claim-evidence-check, sustainability claims are wever_environmental-evidence-check and a warranty claim is wever_warranty-check. Verdict energy_efficiency_inflated, energy_efficiency_partial, energy_efficiency_unverifiable or energy_efficiency_supported. POST /api/energy-efficiency-claim-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_energy-ev-charging-checkAgent Energy: submit one charging session, its posted price and what was charged: the posted price with the station, network, price per kWh, price per minute and whether it runs wh…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,513 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
Description, as the server wrote it Agent Energy: submit one charging session, its posted price and what was charged: the posted price with the station, network, price per kWh, price per minute and whether it runs while charging or while connected, the session fee, the idle fee per minute and its grace minutes, the tax rate, currency, when it was posted and the price as shown; the session with its id, when charging started and stopped, when the vehicle was unplugged, the energy delivered in kWh and its currency; the charges for energy, time, the session fee, idle, tax, any other charge and the total; and optionally the tolerance (0.01 when left out). The signed report gives the session with its charging, connected and idle minutes, each charge beside the posted price times the session’s record as matches, overstated, understated, unposted, within_grace, not_charged or unread with the basis of each, the totals charged, summed and recomputed, and counts them, and names an overstated session: a charge above its recomputed amount, a charge with no posted price, an idle fee inside the grace period and a total above its lines; then a charge below the posted price and a total below its lines, with the fix for each and what would change the verdict. Missing energy or times, no total, no posted price and a currency that differs leave the session unverified. This recomputes one session from its posted price: a fee schedule against a general quote or invoice is wever_fee-evidence-check, fees charged against a stated schedule over many transactions are wever_fee-reconciliation-check and the home rate plan a charger runs on is wever_energy-tariff-check. Verdict energy_charging_overstated, energy_charging_partial, energy_charging_unverified or energy_charging_accurate. POST /api/energy-ev-charging-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_energy-solar-quote-checkAgent Energy: submit a solar or PPA proposal’s claims and the assumptions it states: optionally the proposal with its id, installer, kind (a purchase, a loan, a lease or a power pu…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,846 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
Description, as the server wrote it Agent Energy: submit a solar or PPA proposal’s claims and the assumptions it states: optionally the proposal with its id, installer, kind (a purchase, a loan, a lease or a power purchase agreement) and its sales wording, read for promises such as guaranteed savings or free solar; the system with its size in kW, its claimed first-year production and location; the assumptions with yearly usage, the utility rate, the export credit, the yearly rate rise, yearly degradation, the years counted, the PPA price and its escalator, the lease payment, the system cost, the currency and each incentive with its id, name, kind, amount and whether eligibility was confirmed; the claims of first-year savings, lifetime savings, payback in years and the share of usage covered; and optionally the highest credible yield per kW (1,900 when left out), the highest credible yearly rate rise (4) and the tolerance (5 percent). The signed report gives the system with its production per kW, each claim beside its recomputed figure as within, above_assumptions, rests_on_unconfirmed_incentive or unread, the assumptions with confirmed incentives, the offset and the payback with every incentive counted, and counts them, and names an inflated proposal: production above a credible yield, first-year or lifetime savings above what its own assumptions give, a payback shorter than its cost and confirmed incentives give and an offset above production over usage; then a payback that holds only with unconfirmed incentives, an unconfirmed incentive, a rate rise above the bound, no degradation and promise wording, with the fix for each and what would change the verdict. A missing size, production, rate, price, usage, cost or claim leaves the proposal unverifiable. This recomputes one proposal from its own assumptions: several quotes for one job side by side are wever_quote-compare, one vendor quote is wever_vendor-quote-check and a claim against offered evidence is wever_claim-evidence-check. Verdict energy_solar_inflated, energy_solar_partial, energy_solar_unverifiable or energy_solar_sound. POST /api/energy-solar-quote-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_energy-tariff-checkAgent Energy: submit the rate plan a customer is billed on and the plan they were quoted: optionally the account with its id, utility and customer; the quoted plan with its name, c…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,397 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
Description, as the server wrote it Agent Energy: submit the rate plan a customer is billed on and the plan they were quoted: optionally the account with its id, utility and customer; the quoted plan with its name, code, each rate with its id, component (energy, delivery, demand, a fixed charge or other), time-of-use period, tier, rate and unit, its riders, the dates it takes effect and ends, when it was quoted and where the quote came from; and the billed plan with the same fields and the billing period. The signed report gives the plan on both sides with whether it is the same plan, every rate matched by component, period and tier as matched, above_quote, below_quote, unquoted, not_billed, unrated or unit_differs, the riders quoted, billed, unquoted and not billed, and counts them, and names a mismatch: a billed plan other than the one quoted, a rate above the quote, a charge the quote does not list and a rider the quote does not name; then a rate below the quote, a quoted rate or rider missing from the billed plan, effective dates that differ and a billing period outside the quoted dates, with the fix for each and what would change the verdict. No rates on either side, an unstated rate or unit, no plan identity and no billing period leave the plan unverified. This reads the plan, beside the bill it produces: one bill’s line items recomputed from the tariff and usage are wever_energy-bill-check, a fee schedule against a general quote or invoice is wever_fee-evidence-check and a quote against the governing terms is wever_terms-match-check. Verdict energy_tariff_mismatched, energy_tariff_partial, energy_tariff_unverified or energy_tariff_matched. POST /api/energy-tariff-check. Reads the energy records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Energy reads an energy customer’s own records and recomputes what they claim: the rate plan billed against the plan quoted, a solar or PPA proposal’s production and savings against its stated assumptions, a supply contract against what was agreed, a charging session against the posted price, a carbon credit or certificate claim against its retirement records, a utility bill against its tariff and usage, a commercial demand charge against interval usage and the tariff’s demand rules, and an equipment efficiency or savings claim against its rating and baseline. Beside it, wever_invoice-check reads any invoice against the job it belongs to and wever_fee-evidence-check a fee schedule against a quote or invoice, while an energy check recomputes energy charges from rates, usage and the tariff’s own rules; wever_contract-diff-check compares two versions of one text, while the contract check reads the signed contract against what was agreed; and wever_environmental-evidence-check reads sustainability claims against offered evidence, while the carbon claim check reads the retirement records behind one claim. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_enforcement-layer-checkAgent Access: submit a system’s permission design: each permission with every place it is enforced or stated (gateway, policy engine, platform, credential scope, tool code, databas…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...ated (gateway, policy engine, platform, credential scope, tool code, database, system prom..."
Description, as the server wrote it Agent Access: submit a system’s permission design: each permission with every place it is enforced or stated (gateway, policy engine, platform, credential scope, tool code, database, system prompt, tool description or conversation), what the enforcing layer does when it cannot decide or is down, and whether the model can change that layer through a tool. The signed report names every permission enforced outside the model, every permission that rests only on instructions the model can argue with, ignore, or lose in a long context, or on nothing, and every outside enforcement that fails open or that the model can switch off, with the fix for each. Which tool paths skip the gateway is wever_gateway-coverage-check, the rules the gateway enforces are wever_gateway-rules-check and which decisions code should compute is wever_deterministic-core-check. Verdict prompt_only_permissions, enforcement_gaps, enforcement_unverified or enforced_outside_model. POST /api/enforcement-layer-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_environmental-evidence-checkDiligenceOps: submit environmental or sustainability claims and the evidence offered for them.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit environmental or sustainability claims and the evidence offered for them. The signed report marks each claim supported, contradicted or never mentioned, matching figures with their units, labels such as FSC or LEED Gold and baseline or target years, with quotes. POST /api/environmental-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_escalation-threshold-checkAgent Control: submit an escalation policy for when something is already going wrong: triggers as written or as metric, operator, value and window, each with its action class, owne…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Control: submit an escalation policy for when something is already going wrong: triggers as written or as metric, operator, value and window, each with its action class, owner, response time and when it moves to the next owner, plus the action classes it must cover and the named owners. The signed report names every trigger stated as a judgment rather than a condition an operator could apply at 3 AM, every action class with no named escalation owner and every trigger with no time bound, with the fix for each. Approval before acting is wever_action-threshold-check. Verdict judgment_based, partly_operable or operable. POST /api/escalation-threshold-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_escape-path-checkAgent Control: submit an environment-separation design: sandbox and production credentials (who holds each, whether production accepts it, whether it is predictable), internal tool…Can send data out4 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a URL (input.endpoints): a request to a URL can carry data to whoever runs it.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...paration design: sandbox and production credentials (who holds each, whether production acc..."
- Info
The input schema at required[0] mentions "credentials".
What this check looks forMatched text
inputSchema.required[0]: "credentials"
- Info
The input schema at properties.credentials.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.credentials.description: "Credentials in either environment."
- Info
The input schema at properties.endpoints.items.properties.accepts.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.endpoints.items.properties.accepts.description: "Credentials it accepts; an empty list means none."
Description, as the server wrote it Agent Control: submit an environment-separation design: sandbox and production credentials (who holds each, whether production accepts it, whether it is predictable), internal tools and production endpoints (reach from the sandbox, internet exposure, accepted credentials, bulk access, rate limit) and the sandbox egress default. The signed report names every ordinary path from the sandbox to a production effect (a credential valid in both, a predictable credential, a reachable endpoint that takes no credential or a sandbox one, bulk access with no rate limit) and the smallest change that closes each. What an agent may do inside its environment is wever_authority-scope-check. Verdict escape_path_open, separation_unverified or contained. POST /api/escape-path-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_escrow-briefDiligenceOps: submit one escrow agreement excerpt.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit one escrow agreement excerpt. The signed brief states the parties with their escrow roles, the amount held, the release conditions, how and when funds are released (joint instruction, approval, automatic release, milestones), the return of funds, the dispute path, the agent’s fee and dated deadlines, each quoted, plus what the excerpt leaves unstated. POST /api/escrow-brief. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_escrow-quoteDiligenceOps: quote the exact escrow terms and fees for a shape before anyone commits.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: quote the exact escrow terms and fees for a shape before anyone commits. Inputs: amount_usdc, parties (payer, payee), release_condition (time, oracle_attestation, payer_approval, mutual_approval or delivery_attestation) and an optional expires_at. Returns the Wever rail that fits (conditional payment or managed escrow), its live caps and timing windows, release and refund outcomes, every call with its fee and who makes it, the required fee total and any reason the shape does not fit. The quote moves no money and creates no escrow. POST /api/escrow-quote. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-contract. Human page: https://weverlabs.com/diligence/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-cost-checkAgent Evaluation: submit one evaluation run’s cost and latency figures with its stated budget: the total budget, the per-task cost cap, the task time cap and the percentile it appl…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Evaluation: submit one evaluation run’s cost and latency figures with its stated budget: the total budget, the per-task cost cap, the task time cap and the percentile it applies to (median, 95th or slowest), the run’s task count, total cost and median, 95th percentile and slowest task times, each cost component (agent model calls, judge model calls, tool calls, compute, human review, storage or other) with its cost, and optionally each task’s cost and time. The signed report compares the total, the cost per task and the task time with the budget and caps, adds up the components against the total, and names a total over budget, tasks over the cost cap and task time over its cap, then every component with no cost, components that differ from the total by more than one part in a hundred, tasks with no recorded cost and single tasks over the time cap, with the fix for each. This reads one evaluation run’s cost against its own budget: an agent budget setup’s controls are wever_budget-controls-check, transactions against spending rules are wever_spending-anomaly-check and a forecast of spend is wever_budget-forecast-brief. Verdict over_budget, cost_gaps, cost_unverified or within_budget. POST /api/evaluation-cost-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-dataset-checkAgent Evaluation: submit one evaluation dataset description: its name, size, source (production logs, written by experts, generated, a public benchmark or a mix), version and when …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,277 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
Description, as the server wrote it Agent Evaluation: submit one evaluation dataset description: its name, size, source (production logs, written by experts, generated, a public benchmark or a mix), version and when it was published, the categories the agent must handle beside the categories covered with their example counts and whether hard and edge cases are included on purpose, how the reference labels were made (experts, trained annotators, crowd workers, a model, the agent’s own builders or no one) with the number of labelers, their agreement, a written guide and review of model labels, and the leakage controls: whether the set is held out from training and tuning, whether overlap was checked and found, whether prompt examples come from the set, whether anything was tuned against it, and the model’s training cutoff. The signed report says category by category whether the set covers what the agent must handle, whether the labels can be trusted and whether anything the agent learned from is in the test, and names every required category with no examples, no labels, examples shared with training or tuning data, prompt examples drawn from the set and tuning against the set, then every category under 30 examples, no hard cases, category counts above the dataset size, unreviewed model labels, labels by the agent’s builders, one labeler, agreement under 0.6, no guide, no split, no overlap check and a public benchmark published before the model’s training cutoff, with the fix for each. This reads the set an agent is scored on: whether a benchmark claim’s figures match the source it cites is wever_benchmark-evidence-check and whether a claim is supported by offered evidence is wever_claim-evidence-check. Verdict dataset_unfit, dataset_gaps, dataset_unverified or dataset_sound. POST /api/evaluation-dataset-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-design-checkAgent Evaluation: submit one evaluation plan as supplied: the agent and task, when the plan was written and when the evaluation first ran, the task set with its size and where the …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,183 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
Description, as the server wrote it Agent Evaluation: submit one evaluation plan as supplied: the agent and task, when the plan was written and when the evaluation first ran, the task set with its size and where the tasks come from, each success criterion with the metric that measures it, each metric with what it measures, whether its computation is written down, its pass threshold, whether a pass is at least or at most it and where the threshold comes from (a requirement, the baseline’s score, a prior run, a published benchmark or nowhere), and the baseline with its scores. The signed report says criterion by criterion whether a metric the plan defines measures it, metric by metric whether it has a sourced threshold and a written computation, and whether a baseline is scored, and names no success criteria, every criterion no defined metric measures (quoting wording such as good, reasonable or as expected), an empty task set and criteria written after the first run, then every metric with no threshold, every threshold with no source, every unwritten computation, no baseline, an unscored baseline, an undescribed task set and a task set that is all generated, with the fix for each. This reads the design of one evaluation before it runs: whether a job’s delivered work meets its acceptance criteria is wever_deliverable-check, whether an implementation’s behavior matches its spec is wever_spec-drift-check and whether a benchmark claim’s figures match the source it cites is wever_benchmark-evidence-check. The checklist of a complete evaluation packet is wever_evaluation-snapshot on https://weverlabs.com/mcp-free. Verdict untestable_plan, design_gaps, design_unverified or design_complete. POST /api/evaluation-design-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-drift-checkAgent Evaluation: submit two evaluation runs of the same agent at different times, each with its id, date, model and model version, prompt version, dataset and dataset version, jud…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Evaluation: submit two evaluation runs of the same agent at different times, each with its id, date, model and model version, prompt version, dataset and dataset version, judge and judge version, rubric version, temperature and scores (each with its metric, value, sample size and spread), and optionally the tolerance a score may move by. The signed report compares every condition between the runs, the same, changed or not stated, and every score with how far it moved against the tolerance or, with none, the two runs’ combined noise, and names every changed model, prompt, dataset, judge, rubric or setting that makes the runs measure different things, then every score that drifted, every metric scored in one run only and every score on less than half the earlier sample, with the fix for each. This compares two runs of one agent: whether an implementation’s behavior matches its spec is wever_spec-drift-check, a live endpoint’s reachability, latency and consistency is wever_endpoint-reliability-check and what one run needs to be rerun is wever_evaluation-reproducibility-check. Verdict runs_not_comparable, scores_drifted, drift_unverified or scores_stable. POST /api/evaluation-drift-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-report-checkAgent Evaluation: submit one evaluation summary written for a buyer or principal: its title and audience, each headline score with whether its spread, sample size and a baseline si…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,051 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
Description, as the server wrote it Agent Evaluation: submit one evaluation summary written for a buyer or principal: its title and audience, each headline score with whether its spread, sample size and a baseline sit beside it, the limits with whether they are stated, where they sit (beside the headline, a later section, an appendix or a separate document) and what they are, the failures with whether they are reported, how many tasks failed against how many are reported, whether examples are shown and where they sit, which run conditions are stated (model, dataset, run date, scoring method and settings), and whether the full results are linked, on request or not available. The signed report says whether the limits, failures and run conditions are disclosed alongside the headline scores, and names limits never stated, failures left out or reported in part and no model, dataset or run date stated, then limits or failures placed in an appendix or a separate document, limits not named, headline scores without spread, sample size or baseline, every missing run condition, no failure examples and full results not available, with the fix for each. This reads what an evaluation summary discloses beside its headline: whether AI involvement in content is disclosed is wever_ai-disclosure-check, whether a recommendation discloses its interests is wever_recommendation-disclosure-check and whether a benchmark claim’s figures appear in its cited source is wever_benchmark-evidence-check. Verdict headline_only, disclosure_gaps, disclosure_unverified or fully_disclosed. POST /api/evaluation-report-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-reproducibility-checkAgent Evaluation: submit one reproducibility packet: the agent with its model, exact model version and provider, whether the full prompts are included and their version, the settin…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Evaluation: submit one reproducibility packet: the agent with its model, exact model version and provider, whether the full prompts are included and their version, the settings (temperature, top_p, output limit, seed and others), the tools with whether they are listed and versioned and how outside calls are handled (none, recorded, mocked or live), the dataset with its version, hash and who can get it, the environment with the code version, pinned dependencies and runtime, the scoring method with the scoring script or the judge model and its version, and when it ran. The signed report says essential by essential what someone else needs to rerun the same evaluation and expect the same result, present or missing with why, and names a missing model, a missing or moving model version (such as latest), prompts left out and no dataset version or hash, then missing settings, no seed above temperature 0, unversioned tools, no code version, unpinned dependencies, a missing scoring script or unpinned judge, live outside calls and a private dataset, with the fix for each. This reads what one evaluation needs to be rerun: whether an implementation’s behavior matches its spec is wever_spec-drift-check and two runs of the same agent compared is wever_evaluation-drift-check. Verdict cannot_rerun, reproducibility_gaps, reproducibility_unverified or rerunnable. POST /api/evaluation-reproducibility-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-result-checkAgent Evaluation: submit one evaluation result report: each claim with its text, metric and claimed figure, the subset it comes from and whether the claim names it, the baseline it…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,076 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
Description, as the server wrote it Agent Evaluation: submit one evaluation result report: each claim with its text, metric and claimed figure, the subset it comes from and whether the claim names it, the baseline it compares with and whether it says higher, lower or meets a threshold; each measured result with its metric, value, subset, sample size, spread (a standard deviation, a standard error or a 95 percent interval) and run conditions (model, dataset and date); each baseline result; and the subsets the evaluation ran beside the subsets the report shows. The signed report matches every claim to its measured figure and says whether the figure matches, whether each comparison runs the claimed way and clears the noise in both figures, whether a subset figure says so and whether every subset that ran is reported, and names every claim with no result, every figure the results do not show, every comparison with no baseline result, the wrong way round or inside the noise, every threshold claimed and missed, every undisclosed subset and every dropped subset, then every figure behind a claim with no sample size, fewer than 30 tasks, no spread or no run conditions, with the fix for each. This reads one result report’s claims against its own numbers: whether a benchmark claim’s figures appear in the source it cites is wever_benchmark-evidence-check, whether a claim is supported by offered evidence is wever_claim-evidence-check and one paid call’s result against its order is wever_result-validation-check. Verdict claims_unsupported, result_gaps, result_unverified or claims_supported. POST /api/evaluation-result-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_evaluation-scoring-checkAgent Evaluation: submit one scoring rubric and its judge setup: the rubric version, what it scores, the scale’s lowest and highest scores and step, and each scale point with its a…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,191 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
Description, as the server wrote it Agent Evaluation: submit one scoring rubric and its judge setup: the rubric version, what it scores, the scale’s lowest and highest scores and step, and each scale point with its anchor and whether a worked example is attached; who scores (a model, people, an automatic metric or a mix) with the judge model, whether its version is pinned and its prompt fixed, its temperature, whether it is the model being scored, and its calibration (against human labels, known answers or nothing, on how many items and at what agreement); and the spot check, whether people check a share of the scores, what share, who checks and how often they disagree. The signed report says point by point whether the scale is anchored, whether a model judge is calibrated and pinned, whether a person spot-checks judged scores and whether scores compare across runs, and names an unanchored scale, an uncalibrated model judge, judged scores no person checks and an unpinned judge, then every unanchored point, points off the scale, no worked examples, an unversioned rubric, calibration agreement under 0.7 or on fewer than 50 items, an unfixed judge prompt, a sampling judge, a judge scoring itself, a spot check under one in twenty or by the builders and spot-check disagreement over 0.2, with the fix for each. This reads the rubric and judge behind a score: a trust score computed from verified completed work is wever_agent-trust-scorecard and whether a job’s delivered work meets its acceptance criteria is wever_deliverable-check. A worked rubric is wever_evaluation-rubric-sample on https://weverlabs.com/mcp-free. Verdict scores_unreliable, scoring_gaps, scoring_unverified or scoring_consistent. POST /api/evaluation-scoring-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_exchange-quote-checkAgent Treasury: submit two or more conversion quotes (from and to assets, amounts in and out, any stated rate, fee and fee percentage, and expiry).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Treasury: submit two or more conversion quotes (from and to assets, amounts in and out, any stated rate, fee and fee percentage, and expiry). The signed report computes each effective rate, ranks the quotes, gives the spread between the best and each other quote as a percentage, and flags any quote whose stated fee does not reconcile with its amounts and rate, and any expired quote. Supplied quotes only; live market pricing is wever_market-data-check and wever_fair-value-brief. Verdict ranked, fee_mismatch_found or not_comparable. POST /api/exchange-quote-check. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_expense-close-packCommerce Assurance: submit one period with its payment, receipt and refund records.1 finding
Why it is in this group
- The name "wever_expense-close-pack" says "close": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_expense-close-pack" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_expense-close-pack" says "close": a word for a change.
Description, as the server wrote it Commerce Assurance: submit one period with its payment, receipt and refund records. The signed close pack totals the period by counterparty, by tool or service and by category, each net of the refunds tied to it, lists the exceptions (payments with no receipt, duplicate payments, unmatched refunds, refunds larger than their payment, records outside the period) and closes with a plain summary. Records-based: live wallet activity is wever_wallet-statement. Verdict ready_to_close or exceptions_to_resolve. POST /api/expense-close-pack. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_export-evidence-checkDiligenceOps: submit data export or portability claims about formats, scope and timing and the terms or help pages offered for them.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit data export or portability claims about formats, scope and timing and the terms or help pages offered for them. The signed report marks each claim supported, contradicted or never mentioned, matching formats such as CSV, JSON and XLSX as named items and durations in any unit, with quotes. POST /api/export-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_exposure-cap-checkAgent Governance: submit an exposure or spend cap design: each cap with its amount, what it aggregates over (one order, one agent, one counterparty or the portfolio), its window, w…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Governance: submit an exposure or spend cap design: each cap with its amount, what it aggregates over (one order, one agent, one counterparty or the portfolio), its window, what happens to a request above it (reject, downsize, split, queue or allow), the written rule for a changed request and whether the change is recorded, plus any recorded requests that met a cap. The signed report names every cap with no stated scope or window, every breach left undefined or allowed, every downsizing, split or queue with no written rule or no record of the change, and every recorded request that did not follow its cap. Check order inside one payment flow is wever_firewall-order-check and an approval against a standing limit is wever_approval-policy-check. Verdict cap_undefined, breach_unruled, caps_unverified or caps_defined. POST /api/exposure-cap-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_external-listing-checkAgent Access: submit a listing of your service on someone else’s directory or registry (the directory, the listing page, the endpoint it sends agents to, the network, version, tool…Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a URL (input.listing.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Agent Access: submit a listing of your service on someone else’s directory or registry (the directory, the listing page, the endpoint it sends agents to, the network, version, tool or endpoint count and status, and when it was read) and the live service’s own facts (the endpoints it serves, the endpoints it has retired, the networks it settles on, its version, count and status). The signed report gives every field as match, mismatch or unstated and names every field that would send an agent to the wrong place: a retired or unserved endpoint, a wrong network such as a retired test network, and a live listing for a retired or offline service, then every stale version, count or status, with the correction to request from the directory. Listings on Wever’s own surfaces are wever_listing-check, wever_directory-listing-check and wever_catalog-listing-check, and an offer’s price and terms across every discovery surface are wever_discovery-coverage-check. Verdict misdirects_agents, stale_listing, listing_unverified or matches_live. POST /api/external-listing-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_failure-notice-checkAgent Recovery: submit a failure-handling flow: who the task is done for and how soon they should hear after a task fails for good, each notice sent then with who it goes to and wh…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Recovery: submit a failure-handling flow: who the task is done for and how soon they should hear after a task fails for good, each notice sent then with who it goes to and whether that is the principal, an operator, a team channel or only a log, how soon it goes out and whether it says what failed, what state the work is in and what happens next, plus optionally the tasks that failed for good with when the principal heard and whether from the system, from an operator by hand or by finding out on their own. The signed report says who is told when a task fails for good, how fast and with what detail, and names no notice, notices that reach only operators, channels or logs, and every failure the principal found out about on their own, then every notice past the target, every notice missing what failed, the state of the work or what happens next, and every principal told by hand or late, with the fix for each. This reads the notice to the principal: who owns the response to an incident is wever_incident-ownership-check, an incident response plan’s notification windows are wever_incident-plan-check and where a refused or failed handed-off task goes is wever_handoff-failure-check. Verdict principal_finds_out, notice_gaps, notice_unverified or principal_told. POST /api/failure-notice-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_fair-value-briefWever Briefs: the deep brief behind the free fair-value lookup for one prediction market.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Wever Briefs: the deep brief behind the free fair-value lookup for one prediction market. Inputs: platform (kalshi or polymarket) and market (ticker, slug or market URL). Returns the estimated fair price, every input used (engine estimate, compared price, Kalshi, Polymarket and Novig quotes where available, and a direct venue read), the spread in points against each quoted price and a reasoning summary, with the engine’s own reasoning when it produced this exact estimate. One market per call. POST /api/fair-value-brief. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns one brief with an Ed25519 signed receipt. Contract: /api/briefs-contract. Human page: https://weverlabs.com/briefs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_fee-evidence-checkDiligenceOps: submit a fee schedule and a quote or invoice, as text or as lines.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a fee schedule and a quote or invoice, as text or as lines. The signed report pairs each charge with its schedule fee by label, lists the charges the schedule does not list and the listed fees charged at a different amount, rate or currency, with the expected amount and the difference to the cent. POST /api/fee-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_fee-reconciliation-checkAgent Treasury: submit the fees you were charged (date, payer or payee, amount, fee label and the transaction amount each was charged on) with the stated fee schedule (percentage r…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Treasury: submit the fees you were charged (date, payer or payee, amount, fee label and the transaction amount each was charged on) with the stated fee schedule (percentage rates, flat fees, minimums and caps). The signed report recomputes each expected fee and itemizes overcharges, undercharges and fees with no matching schedule entry, with totals. Public fee claims are wever_fee-evidence-check. Verdict reconciled or discrepancies_found. POST /api/fee-reconciliation-check. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-actionability-checkAgent Feedback: submit one piece of feedback on a specific piece of work, point by point: optionally the giver and the work, and each point with its id, the point as written, its t…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,484 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
Description, as the server wrote it Agent Feedback: submit one piece of feedback on a specific piece of work, point by point: optionally the giver and the work, and each point with its id, the point as written, its tone (praise, criticism, a suggestion, a question or mixed, read from wording such as great job or confusing when left out), the specific part of the work it is about, what it says is wrong or right there, the change it states or implies, whether it is about something outside the receiver’s control, whether it presents that as the receiver’s own defect, and the ids of other points whose instruction it contradicts. The signed report gives every point as actionable, nothing_to_act_on, misplaced_blame, conflicting, unlocated, no_change or unconfirmed, with every pair of contradicting instructions, and names feedback that leaves the receiver nothing clear to act on: every point pure praise or pure complaint with no part of the work and no change named, something outside the receiver’s control presented as their defect, and two instructions in the same feedback that contradict each other; then some such points among actionable ones, a change with no part named and a criticism of a part with no change, with the fix for each. This reads whether a receiver can act on feedback, not scoring or benchmarking: grading an agent’s output against a rubric is wever_evaluation-scoring-check, and a collision between agents’ recorded actions on shared work is wever_coordination-conflict-check. Verdict feedback_unactionable, feedback_actionability_gaps, feedback_actionability_unverified or feedback_actionable. POST /api/feedback-actionability-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-closure-checkAgent Feedback: submit the loop records for feedback on a specific piece of work: each loop with its id, what the feedback was about, the giver, the resolution state the record hol…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,386 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
Description, as the server wrote it Agent Feedback: submit the loop records for feedback on a specific piece of work: each loop with its id, what the feedback was about, the giver, the resolution state the record holds (done, declined, deferred or open), the reason for a decline or deferral, the date a deferred loop comes back, when the record marks it closed, whether the giver was told the outcome and when, whether the feedback reported a defect in the work, and whether someone verified the fix and who; and optionally the moment to read the loops at. The signed report gives every loop as closed, no_state, closed_untold, fix_unverified, open, declined_unexplained, deferred_undated, deferral_lapsed or unconfirmed, counts the loops by state with the givers told, gives the defect reports with their verification, and names a feedback loop left open: loops with no recorded state at all, closure claimed in the record but never communicated to the giver, and defect reports closed as done with nobody verifying the fix; then loops still marked open, declines with no reason, deferrals with no date and deferrals past their date, with the fix for each. This reads the working loop between one giver and one receiver: how a group of agents escalates and resolves a disagreement is wever_coordination-escalation-check, and an operator’s continuity arrangements for keys, successors and standing jobs are wever_continuity-check. Verdict feedback_loop_open, feedback_closure_gaps, feedback_closure_unverified or feedback_loop_closed. POST /api/feedback-closure-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-freshness-checkAgent Feedback: submit the feedback being relied on now about a specific piece of work: optionally the work with its current version and each version with when it was released and …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,481 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
Description, as the server wrote it Agent Feedback: submit the feedback being relied on now about a specific piece of work: optionally the work with its current version and each version with when it was released and whether it is still in use; each piece of feedback with its id, the point as given, the version it is about, when it was given, its tone, how it is relied on now (as evidence of quality, as the basis for a change, as an open issue or other), the id of later feedback on the same point that supersedes it and a later change that already answers it; the moment it is relied on; and how many days old praise may be and still count as current evidence of quality (180 when left out). The signed report gives the work and its versions, every piece of feedback as current, version_gone, stale_praise, superseded, answered, earlier_version or unconfirmed with its age in days, and the oldest and newest ages, and names feedback that no longer describes the work: acting today on feedback about a version no longer in use, and old praise presented as current evidence of quality; then feedback superseded by later feedback on the same point, feedback a later change already answers, and feedback about an earlier version with no record of whether it still applies, with the fix for each. How an agent’s durable reputation score ages is wever_reputation-decay-check, drift in an evaluation’s results is wever_evaluation-drift-check, and whether an agent’s knowledge entries are current is wever_knowledge-freshness-check. Verdict feedback_stale, feedback_freshness_gaps, feedback_freshness_unverified or feedback_current. POST /api/feedback-freshness-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-gaming-checkAgent Feedback: submit one set of feedback about one piece of work or one agent: optionally the work and the receiver; each entry with its id, the giver, when it was given, its ton…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,557 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
Description, as the server wrote it Agent Feedback: submit one set of feedback about one piece of work or one agent: optionally the work and the receiver; each entry with its id, the giver, when it was given, its tone (positive, negative, mixed or neutral), whether the giver is related to the receiver (the same operator, owner or team, or a paid affiliate), the other givers in the set the giver is related to, whether the receiver asked for it and the giver’s connection to the work on record (used it, saw it or none); the disputes between the receiver and a giver with when each was raised or decided; and who the receiver asked for feedback, which of them were expected to be positive and how they were chosen. The signed report gives every entry as clean, flagged, set_apart or unconfirmed, each manipulation pattern with the entries behind it, and counts the positive, negative, related, unconnected and solicited entries, and names manipulation of the exchange itself: coordinated praise from two or more related parties, negative feedback filed within 14 days after a dispute with the receiver, feedback solicited only from parties expected to be positive, and three or more entries from accounts with no connection to the work within 72 hours; then a single related-party entry, a scattered unconnected entry and a hand-picked selection, with the fix for each. This grades one specific feedback set: whether a reputation system’s design resists manipulation of its durable standing record is wever_reputation-gaming-check, and a dispute over a reputation record is wever_reputation-dispute-check. Verdict feedback_set_manipulated, feedback_set_gaps, feedback_set_unverified or feedback_set_clean. POST /api/feedback-gaming-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-incorporation-checkAgent Feedback: submit the receiver’s response and revision record for one round of feedback on a specific piece of work: optionally the reply to the giver with its text and when i…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,470 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
Description, as the server wrote it Agent Feedback: submit the receiver’s response and revision record for one round of feedback on a specific piece of work: optionally the reply to the giver with its text and when it was sent, read for acknowledgement-only wording such as thanks, noted; the revised version with when it was made; and each feedback item with its id, the point as given, whether it was acknowledged to the giver, what the receiver did (changed the work, declined, deferred or nothing), the reason for a decline or deferral, the date a deferred item will be taken up, the change the reply claims for it and whether the revised work holds that change. The signed report gives every item as incorporated, declined_with_reason, deferred_with_date, answered_with_reason, noted_only, claimed_absent, unacknowledged, declined_unreasoned, deferred_undated or unconfirmed, counts the acknowledged, changed, declined and deferred items and the changes present in the revision, and names feedback that did not land: items never acknowledged at all, thanks, noted with no change and no reason, and changes the reply claims that the revised work does not hold; then declines with no reason and deferrals with no date, with the fix for each. This reads one round of response to feedback, not a standing record: an agent’s durable record of completed jobs is wever_reputation-record-check, and a collision between agents’ recorded decisions on shared work is wever_coordination-conflict-check. Verdict feedback_not_landed, feedback_incorporation_gaps, feedback_incorporation_unverified or feedback_incorporated. POST /api/feedback-incorporation-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-provenance-checkAgent Feedback: submit who each piece of feedback in one exchange came from: optionally the work and its current version, and each piece of feedback with its id, the giver as recor…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,600 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
Description, as the server wrote it Agent Feedback: submit who each piece of feedback in one exchange came from: optionally the work and its current version, and each piece of feedback with its id, the giver as recorded (left out when unattributed), how it is treated (as authoritative, as one input or set aside), whether the giver saw or used the actual work, the version it judges and the versions the giver saw, whether the giver was asked for it, whether the giver holds a role that authorizes them to judge the work, and any stake in the outcome (none, a competitor, a counterparty in a dispute, a vendor, related to the receiver or other) with whether it is disclosed. The signed report gives every piece of feedback with its standing as sound, unattributed_authoritative, unseen, interest_undisclosed, unattributed, unsolicited or unconfirmed, counts the attributed, seen, asked or authorized and disclosed, and names feedback carrying weight its source does not support: unattributed feedback treated as authoritative, feedback about work or a version of it the giver never saw, and an undisclosed stake such as a competitor or a counterparty in a dispute behind a judgment; then unattributed feedback weighed as one input and givers neither asked nor authorized, with the fix for each. These provenance and consent questions are about one feedback exchange: where a fact in an agent’s stored memory came from is wever_memory-provenance-check, whether a principal agreed to be remembered is wever_memory-consent-check, and the sources behind an agent’s durable reputation record are wever_reputation-source-check. Verdict feedback_provenance_unsound, feedback_provenance_gaps, feedback_provenance_unverified or feedback_provenance_sound. POST /api/feedback-provenance-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_feedback-report-checkAgent Feedback: submit the feedback summary written for the principal, with the records behind it: its title, its audience, its headline as written and optionally its text; whether…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,533 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
Description, as the server wrote it Agent Feedback: submit the feedback summary written for the principal, with the records behind it: its title, its audience, its headline as written and optionally its text; whether it states the negative feedback, the declined items (with how many carry their reasons), the open loops (and whether it presents them as resolved) and the gaming flags found in the set, where each sits (beside the headline, in a later section, in an appendix or in a separate document) and how many it names; whether it presents an average and whether the disagreement underneath sits beside it; and optionally the record of negative items, declined items, open loops, gaming flags and points on which givers disagree. The signed report gives the summary with any praise-only wording, the disclosure of each kind of record and of the average, and the record, and names a summary that whitewashes the feedback: negative feedback left out or stated as none while the record holds some, so it reports only praise; open loops left out or presented as resolved; gaming flags left out; and an average presented without the disagreements underneath it; then declined items left out or without their reasons, counts below the record’s, disclosures in an appendix or a separate document, and wording such as great feedback or overwhelmingly positive while the record holds negative feedback, with the fix for each. An agent’s durable reputation summary is wever_reputation-report-check, and an evaluation summary of scores is wever_evaluation-report-check. Verdict feedback_summary_whitewashed, feedback_summary_gaps, feedback_summary_unverified or feedback_summary_faithful. POST /api/feedback-report-check. Reads only the recorded feedback material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no feedback is sent, no party is contacted and no work is changed. Agent Feedback reads the working feedback loop on one piece of work between a giver and a receiver: it is not scoring or benchmarking (the wever_evaluation checks), not a durable standing record (the wever_reputation checks) and not how agents organize shared work (the wever_coordination checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_firewall-order-checkAgent Control: submit a payment firewall’s design: its checks with where each runs (before the order, before settlement, after settlement or alongside it) and what a failure does, …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Control: submit a payment firewall’s design: its checks with where each runs (before the order, before settlement, after settlement or alongside it) and what a failure does, its cap with its scope and what counts toward it, the paths an order takes to settlement (retry, split, reorder, manual) and what it records. The signed report names every check that runs after settlement instead of before it, every path that bypasses the cap, and whether passes and rejects are both recorded with the order, size, check, result, reason and time. Verdict settles_unchecked, enforcement_unverified, record_incomplete or enforced. POST /api/firewall-order-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_fleet-roster-checkAgent Handoffs: submit the lists of agents an organisation keeps (each with its keeper, whether the organisation names it the roster, how it is kept current, when it was last updat…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...registry covers what deployment, usage, credential and billing lists show is wever_agent-r..."
Description, as the server wrote it Agent Handoffs: submit the lists of agents an organisation keeps (each with its keeper, whether the organisation names it the roster, how it is kept current, when it was last updated and the agents it lists with each owner), the agents added, changed or retired with when the roster showed each change, and how soon a change should reach the roster. The signed report says whether a single roster exists and names no list at all, several lists with none or more than one named the roster, and an added agent the roster does not show (a fleet known only by anecdote), then every agent known only to a side list, every agent with no owner, a roster with no keeper or no update practice, every change the roster never showed or showed late and every retired agent still listed, with the fix for each. This reads whether the organisation keeps one roster and keeps it current as agents come and go: whether one registry covers what deployment, usage, credential and billing lists show is wever_agent-registry-check, and the outside surfaces where buying agents find a seller are wever_discovery-coverage-check. A summary sample is wever_fleet-roster-sample on https://weverlabs.com/mcp-free. Verdict fleet_known_anecdotally, roster_gaps, roster_unverified or single_current_roster. POST /api/fleet-roster-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_gateway-coverage-checkAgent Accountability: submit an MCP or tool-access architecture: the gateway with the checks it runs, what decides allow or deny and what happens when it is down, and every tool wi…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Accountability: submit an MCP or tool-access architecture: the gateway with the checks it runs, what decides allow or deny and what happens when it is down, and every tool with its effect, whether it is reached through the gateway, who reaches it directly and where its calls are traced. The signed report names every tool path that skips the gateway’s deterministic checks, every tool reachable both through the gateway and directly, and whether tracing covers the unmediated paths. Check order inside one payment flow is wever_firewall-order-check. Verdict unmediated_paths, coverage_unverified or fully_mediated. POST /api/gateway-coverage-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_gateway-rules-checkAgent Access: submit the rules a gateway or policy layer enforces, each with where it is written (policy document, decision record, versioned or unversioned configuration, code onl…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Access: submit the rules a gateway or policy layer enforces, each with where it is written (policy document, decision record, versioned or unversioned configuration, code only or nowhere) and its source, its owner and owner type, when it last changed and was last reviewed, whether it is reviewed when the toolset changes and the tools it covers, plus the toolset with when it last changed and the tools it serves now. The signed report gives each rule written, owned and current, and names every rule written nowhere or held only in live configuration or code, owned by no one, by an agent or only as a vendor default, reviewed before the last toolset change or still naming retired tools, and not reviewed when the toolset changes. Which routes the gateway reaches is wever_gateway-coverage-check and which instruction governs one action is wever_policy-precedence-check. A summary sample is wever_gateway-rules-sample on https://weverlabs.com/mcp-free. Verdict unwritten_rules, rule_gaps, rules_unverified or rules_governed. POST /api/gateway-rules-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-alignment-checkAgent Goals: submit the working goal with what the principal asked for: the ask as the principal gave it, with whether it is the principal’s own words, the originating instruction …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,836 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit the working goal with what the principal asked for: the ask as the principal gave it, with whether it is the principal’s own words, the originating instruction or a restatement, and when it was given; the goal with its id and outcome; and each requirement in the ask with its id, the requirement as the principal stated it, the same requirement as the goal records it and who approved any change to it. The signed report gives the ask, the goal, every requirement with whether it is quoted from the ask and its status as kept, reworded, changed, dropped, approved_change or unconfirmed, and counts the requirements kept, reworded, changed, dropped and changed with approval, and names a goal that no longer matches the ask: requirements dropped from the goal and requirements carried with another figure than the principal stated, with no approval from anyone but the agent itself; then requirements carried in other words, requirements not found in the ask as supplied and changes approved by a team or TBD rather than a named principal, with the fix for each. A goal read only against a restatement is not read against the principal’s own words. This reads intent, not permission: whether an actor stayed within its granted authority is wever_authority-scope-check, the mandate tools such as wever_mandate-issue record authorization to act, one negotiation inside its mandate is wever_negotiation-mandate-check, and a purchase intent quoted for approval is wever_purchase-create-intent. Verdict goal_misaligned, alignment_partial, alignment_strained or goal_aligned. POST /api/goal-alignment-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-conflict-checkAgent Goals: submit one active goal with the principal’s other active goals, standing constraints and limits, and the comparisons made: the goal with its id and outcome; each other…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,807 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit one active goal with the principal’s other active goals, standing constraints and limits, and the comparisons made: the goal with its id and outcome; each other goal, constraint or limit with its id, its kind and the statement as the principal gave it; and each comparison with the id it compares, whether the two contradict, what contradicts what, how it was settled, read for averaging wording such as balance, compromise or split the difference, and who settled it (the principal, the agent or another party). The signed report gives every goal, constraint and limit with its comparison and its status as clear, open, averaged, traded, agent_decided, settled, unexamined or unconfirmed, and counts the comparisons, the contradictions and those open, averaged and settled, and names a goal in unsettled conflict: contradictions left open, contradictions averaged away by anyone but the principal, and constraints or limits traded against the goal by the agent alone; then contradictions settled and kept on the record, choices between two of the principal’s goals made by the agent, and conflicts with no detail or no recorded decider, with the fix for each. Each contradiction is named rather than averaged away. This reads one goal against everything else the principal wants: collisions between collaborators’ actions on one shared goal are wever_coordination-conflict-check, and the plan for dividing the work is wever_coordination-plan-check. Verdict goal_in_conflict, conflicts_noted, conflict_unexamined or conflicts_checked_clean. POST /api/goal-conflict-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-drift-checkAgent Goals: submit a goal with the agent’s recorded actions over time: the goal with its id, outcome and scope; each action with its id, when it was taken, what the agent did, wha…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,561 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit a goal with the agent’s recorded actions over time: the goal with its id, outcome and scope; each action with its id, when it was taken, what the agent did, what it was for (the outcome, a scope item or something else), the reason when it was for something else and who approved it; and optionally how many consecutive off-goal actions count as drift (3 when left out). The signed report gives every action in time order with its status as on_goal, approved_detour, explained_detour, off_goal or unlinked, counts the actions on the goal, the approved detours, the unapproved off-goal actions and the longest run, and states the earliest divergence: the first unapproved action away from the goal, when it came and how many on-goal actions came before it, and names drift: a run of consecutive actions that serve something other than the goal with no approval; then single off-goal actions with no reason and detours with a reason but no approval, with the fix for each. This reads an agent’s actions drifting away from its principal’s stated objective: drift in evaluation scores and results over time is wever_evaluation-drift-check, and a build drifting from its specification is wever_spec-drift-check; neither is goal drift. Verdict goal_drifted, drift_minor, drift_unmeasured or drift_absent. POST /api/goal-drift-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-freshness-checkAgent Goals: submit a goal with what it rests on and what came after it: the goal with its id, outcome, when it was set, when its context expires, when it was last reviewed and how…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,747 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit a goal with what it rests on and what came after it: the goal with its id, outcome, when it was set, when its context expires, when it was last reviewed and how often it is reviewed; each premise with its id, what the goal assumes, whether it still holds and when it was last checked; each instruction received since with its id, when it came, whether it came from the principal, its text, read for wording such as instead, no longer or change of plan, and whether it replaces or ends the goal; and optionally how old a premise check may be (30 days when left out) and the moment to read the goal at. The signed report gives the goal, every premise with its age and its status as holds, stale, failed, undated or unconfirmed, every instruction with any replacing wording and its status as supersedes, weighed, unweighed, before_goal, not_principal or unconfirmed, and counts the premises holding, failed and stale and the instructions superseding and unweighed, and names a goal overtaken: a later instruction from the principal that replaces it, a premise that no longer holds and a context that has expired; then later instructions with replacing wording never weighed against the goal, premises checked too long ago and a review overdue, with the fix for each. This reads whether the objective itself is still current: the age of the facts inside one answer is wever_uncertainty-freshness-check. Verdict goal_superseded, freshness_partial, goal_possibly_stale or goal_current. POST /api/goal-freshness-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-progress-checkAgent Goals: submit the progress claimed toward a goal with the records behind it: the goal with its id, outcome and success test; each claim with its id, the progress claimed, rea…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,700 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit the progress claimed toward a goal with the records behind it: the goal with its id, outcome and success test; each claim with its id, the progress claimed, read for wording such as almost done or on track, what it was measured by, the figure claimed and the ids of the records behind it; and each record with its id, what it is (a test run, a metric, a commit or a file) and the figure it holds. The signed report gives the goal, every claim with whether its measure is the success test, the records it cites, the highest figure they hold, any asserted wording and its status as evidenced, above_record, claimed_only, off_test, unmeasured, unbacked or unconfirmed, and counts the claims evidenced, claimed only, above their records, measured off the test and unbacked, and names progress claimed rather than shown: claims with no measure and no record, and claims above the figures their records hold; then claims measured by something other than the success test, claims citing records with no measure and claims with no record, with the fix for each. This reads progress toward a principal’s objective against the goal’s own success test: a group’s progress against its shared plan is wever_coordination-progress-check, a job’s recorded outcome is wever_trace-outcome-brief and an open promise’s status is wever_commitment-tracking-check. Verdict progress_claimed_only, progress_partial, progress_unmeasured or progress_evidenced. POST /api/goal-progress-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-report-checkAgent Goals: submit the goal report sent to the principal, with the goal record behind it: its title, the goal it reports on, the period it covers and its text, read for wording su…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,699 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit the goal report sent to the principal, with the goal record behind it: its title, the goal it reports on, the period it covers and its text, read for wording such as on track, no blockers or nearly done; for the goal’s standing (met, on track, at risk or off track), its progress as a percent of the success test, its open blockers and the changes made to it, whether the report states each, what it states and whether each links to the goal record behind it; and optionally the goal record’s standing, progress, open blockers and changes. The signed report gives the report with any wording that calls the goal safe, each part with whether it is stated, what it states and whether it is backed by the record, and the record, and names a report that flatters the goal record: a better standing, more progress, fewer blockers or fewer changes than the record shows, parts with no link to the record, and wording such as on track or no blockers while the record shows the goal at risk, off track or blocked; then parts left out, parts misstated the other way and a report that does not name its goal, with the fix for each. This reports one goal to its principal against the goal record: a kept-promises report is wever_commitment-report-check, a learning report is wever_learning-report-check and a triage report is wever_triage-report-check. Verdict goal_report_inflated, goal_report_partial, goal_report_missing or goal_report_sound. POST /api/goal-report-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-scope-checkAgent Goals: submit a goal’s working scope with the scope the principal authorized: each authorized item as the principal recorded it, each item the principal ruled out, and each w…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,455 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit a goal’s working scope with the scope the principal authorized: each authorized item as the principal recorded it, each item the principal ruled out, and each working scope item with its id, the item, when it joined the working scope and, for an addition, who approved it and when. The signed report gives every authorized item with whether it is worked on, every working item with the authorized item it matches and its status as authorized, approved_addition, unapproved, excluded, unconfirmed or unread, and counts the items authorized, added with approval, unapproved and ruled out, and names a working scope grown past what the principal authorized: additions with no approval, or approved only by the agent itself, and items the principal ruled out; then additions approved by a team or TBD rather than a named principal and additions that joined the working scope before their approval, with the fix for each. This reads intent, not permission: whether an actor stayed within its granted authority is wever_authority-scope-check, and the written bounds of one delegation are wever_delegation-scope-check. Verdict scope_exceeded, scope_partial, scope_unrecorded or scope_as_authorized. POST /api/goal-scope-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_goal-statement-checkAgent Goals: submit an agent’s goals as recorded: each goal with its id, the outcome wanted, read for activity wording such as improve, optimize or work on, what it covers and what…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,652 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
Description, as the server wrote it Agent Goals: submit an agent’s goals as recorded: each goal with its id, the outcome wanted, read for activity wording such as improve, optimize or work on, what it covers and what it leaves out, how to tell it is met, read for opinion wording such as looks good or everyone is happy, when it is due and when it is next reviewed. The signed report gives every goal with the parts it is missing, any vague wording read from it, whether its review comes only after its horizon and its status as complete, partial, vague or unstated, counts the goals complete, partial, vague and unstated, and lists every vague word found, and names goals stated too vaguely to act on: outcomes worded as an activity with no figure or date, and success tests that leave success to opinion; then goals with no scope, no success test or no horizon or review point, and reviews set only after the horizon, with the fix for each. A goal with no outcome is not stated. This reads the outcome an agent works toward, recorded before any promise exists: a promise once made is wever_commitment-specificity-check, the success criteria of an evaluation plan are wever_evaluation-design-check, one job’s objective in a resume pack is wever_job-handoff-pack and whether a handoff format carries the goal to the next agent is wever_context-transfer-check. Verdict statement_vague, statement_partial, goal_unstated or statement_complete. POST /api/goal-statement-check. Reads only the recorded goal material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no goal is set, changed, approved or retired and no plan is changed. Agent Goals reads an agent’s goal: whether it is well formed, whether it still matches what its principal actually asked for and whether the work has drifted from it. It is not a promise (promises and deadlines once they are made, the wever_commitment checks; a goal is the intended outcome those promises serve, recorded before any promise exists), not how work is divided (the wever_coordination checks, where wever_coordination-plan-check reads the plan for dividing the work; the goal checks read the objective that plan is supposed to serve), not authorization (wever_mandate-issue, wever_mandate-status, wever_mandate-amend, wever_mandate-halt, wever_mandate-revoke, wever_mandate-draft and wever_ap2-mandate-gateway record authorization to act, wever_negotiation-mandate-check reads one negotiation inside its mandate and wever_authority-scope-check reads an actor inside its granted authority; a goal records the outcome the agent acts toward, a question of intent rather than permission, and neither stands in for the other), not drift in scores or builds (wever_evaluation-drift-check reads evaluation scores and results drifting over time and wever_spec-drift-check a build drifting from its specification; neither is goal drift), not confidence in information (the wever_uncertainty checks; the goal checks read intent and fit), not learning (the wever_learning checks read an agent updating its methods from experience, and methods may change freely; a goal changes only when the principal’s ask changed or the principal approved the change) and not the dates inside one document (wever_deadline-check; the goal checks read a goal and its work over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_guarantee-evidence-checkDiligenceOps: submit a guarantee as promised (money-back, performance or satisfaction) and the guarantee terms.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a guarantee as promised (money-back, performance or satisfaction) and the guarantee terms. The signed report lists every deadline, condition and exclusion in the terms that differs from the promise: a shorter or differently counted window, named conditions such as proof of purchase or original packaging, exclusions, fees deducted from the refund and credit in place of money, each quoted. POST /api/guarantee-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_handoff-acceptance-checkAgent Handoffs: submit a handoff flow: each sender-to-receiver route with how the receiver confirms it has taken the task (an explicit acceptance, only a delivery receipt from the …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Handoffs: submit a handoff flow: each sender-to-receiver route with how the receiver confirms it has taken the task (an explicit acceptance, only a delivery receipt from the transport, or nothing), when the sender lets go (on acceptance or on send), how long the sender waits for acceptance and whether each acceptance is recorded, plus optionally the transfers made on those routes with when each was sent, accepted and released. The signed report names every route where the sender lets go on send or the receiver confirms nothing, every route that takes a delivery receipt for acceptance, has no acceptance window or keeps no record of acceptance, and every transfer let go with no acceptance, let go before its acceptance or accepted after its window, with the fix for each. This reads who holds a task between send and acceptance: what happens after a receiver refuses, fails or goes quiet is wever_handoff-failure-check, a delivered result against its order is wever_result-validation-check and a job’s work against its acceptance criteria is wever_deliverable-check. Verdict work_can_drop, acceptance_gaps, acceptance_unverified or held_until_accepted. POST /api/handoff-acceptance-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_handoff-failure-checkAgent Handoffs: submit a handoff flow with its failure paths: for each route, where the task goes when the receiver refuses it, fails mid-task or goes quiet (back to the sender, to…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Handoffs: submit a handoff flow with its failure paths: for each route, where the task goes when the receiver refuses it, fails mid-task or goes quiet (back to the sender, to a named fallback, escalated, retried or dropped), how long without word counts as silence, the fallback and the escalation recipient with whether that is a named person, a role, a channel or an agent, when the path was written down and when a receiver on the route first refused, failed or went quiet. The signed report follows each failure to its end and names every route with no path for a refusal, a failure or silence, every dropped task, every fallback or escalation to no one, every silence path with no time after which quiet counts, every escalation to a role, channel or agent rather than a named person, every path that only retries and every path written down after the first failure, with the fix for each. This reads the road back for a handed-off task: where work goes when it hits a limit is wever_breach-path-check, an operator’s fallback when an agent is wrong or down is wever_incident-ownership-check and an approver who is unavailable is wever_approval-chain-check. Verdict no_road_back, failure_path_gaps, failure_path_unverified or road_back_defined. POST /api/handoff-failure-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_handoff-record-checkAgent Handoffs: submit a system’s task handoff records (each with its id, the task, who sent it, who received it, when, where the task stood when it moved, the constraints it trave…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Handoffs: submit a system’s task handoff records (each with its id, the task, who sent it, who received it, when, where the task stood when it moved, the constraints it travels under such as a deadline, budget or limit, and when the record was written) and optionally the task moves seen in other logs (a tracker, a queue or message logs). The signed report counts every handoff that keeps the essentials and names every record missing the task, its current state or who sent it (the core a handoff cannot be reconstructed without), every record missing the constraints, every record written more than an hour after the handoff, and every move seen elsewhere with no handoff record behind it, with the fix for each. This reads the records of handoffs a system made: one job’s state assembled into a resume pack is wever_job-handoff-pack, a signed sender-to-recipient pack is wever_agent-to-agent-handoff-pack, an operator handover package is wever_handover-check and the format every handoff uses is wever_context-transfer-check. Verdict handoffs_are_rumors, record_gaps, records_unverified or handoffs_recorded. POST /api/handoff-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_handoff-scope-checkAgent Handoffs: submit a delegation setup: optionally the sending agent and its own access (tools, data and spending), and each task it hands to another agent with what the task ne…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "..., the receiver’s own standing access, a credential several agents share or the sender’s ow..."
- Info
The input schema at properties.delegations.items.properties.grant.properties.via.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.delegations.items.properties.grant.properties.via.description: "..., the receiver’s own standing access, a credential several agents share, or the sender’s o..."
Description, as the server wrote it Agent Handoffs: submit a delegation setup: optionally the sending agent and its own access (tools, data and spending), and each task it hands to another agent with what the task needs (tools, data and spending) and what the receiver gets: its tools, data and spending, how the access reaches it (a grant made for the task, the receiver’s own standing access, a credential several agents share or the sender’s own credential), when it ends and whether the receiver can pass it on. The signed report names every handoff where the sender’s access travels with the task, through the sender’s own credential or a grant copying the sender’s whole scope, then every tool, data set and spending amount granted beyond what the task needs, every shared credential, every need the grant leaves out, every grant that outlives the task and every grant the receiver can pass on, with the fix for each. This reads agent-to-agent task delegation: a buyer acting for a human principal is wever_delegated-identity-check, an agent meant to change nothing is wever_read-only-scope-check and one agent’s access against its own authority grant is wever_authority-scope-check. Verdict sender_access_travels, scope_gaps, scope_unverified or scoped_to_task. POST /api/handoff-scope-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_handover-checkAgent preflight round 2: check a handover package before a receiver takes over.2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...eceiver takes over. Input text, fields (credentials, runbooks, billing, domains, data_expor..."
- Info
The input schema at properties.fields.properties.credentials.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.fields.properties.credentials.description: "How credentials and key custody pass to the receiver."
Description, as the server wrote it Agent preflight round 2: check a handover package before a receiver takes over. Input text, fields (credentials, runbooks, billing, domains, data_export, obligations) or both. The signed report marks credentials and custody transfer, runbooks, billing ownership, domains and DNS, data export and open obligations covered, partial or not covered, with quotes; states what a receiver could and could not run on day one; and names each gap. POST /api/handover-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_identity-portability-checkAgent Accountability: submit an agent identity setup: the issuer, the kind of identity, who holds its keys, how another party verifies it, its export, who can revoke it, the platfo…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Accountability: submit an agent identity setup: the issuer, the kind of identity, who holds its keys, how another party verifies it, its export, who can revoke it, the platforms that accept it and what it carries, such as reputation or payment history, with what each is tied to. The signed report says whether the identity can be presented and verified outside the issuing platform, what breaks if the issuer revokes it or the platform closes the door, and every capability tied to a platform-kept identity with no export path. Identity record consistency is wever_agent-identity-check. Verdict platform_locked, partly_portable or portable. POST /api/identity-portability-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_incident-evidence-checkDiligenceOps: submit incident claims about duration, scope, cause and resolution and the timeline, status updates or postmortem offered for them.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit incident claims about duration, scope, cause and resolution and the timeline, status updates or postmortem offered for them. The signed report marks each claim supported, contradicted or never mentioned, comparing durations in any unit, clock times, regions and affected shares, and checks a stated duration and start or resolution time against the timeline the evidence gives, with quotes. POST /api/incident-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_incident-ownership-checkAgent Governance: submit an agent operations setup: each agent with whether it runs in production, its named owner and owner type, who responds and who is paged when it errs, and i…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Governance: submit an agent operations setup: each agent with whether it runs in production, its named owner and owner type, who responds and who is paged when it errs, and its fallback for when it is wrong or down (manual process, previous version, kill switch or degraded mode) with who runs it and whether it has been exercised. The signed report names every production agent with no named person as owner, every error path with no responder or no one paged, and every agent with no fallback or an unowned or untested one. Escalation timing is wever_escalation-threshold-check, the incident plan document is wever_incident-plan-check and evidence after an incident is wever_incident-evidence-check. Verdict unowned_agents, response_gaps, ownership_unverified or owned_before_incident. POST /api/incident-ownership-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_incident-plan-checkAgent preflight round 3: check your own incident response plan before you depend on it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight round 3: check your own incident response plan before you depend on it. Input text, fields (detection, severity_levels, notification_windows, roles, customer_communication, postmortem, payment_incidents) or both. The signed report marks each area covered, partial or not covered, with quotes, the severity levels and notification windows it reads, and names each gap plainly: no detection path, no notification window, no named owner, no customer communication, no postmortem practice and no path for an incident that involves payments. To check a vendor’s past incident record against its claims, use wever_incident-evidence-check. POST /api/incident-plan-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-certificate-checkAgent Insurance: submit a certificate of insurance with the policy it was issued from: the certificate with the date it was issued, the insurer, the policy number, the named insure…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,131 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
Description, as the server wrote it Agent Insurance: submit a certificate of insurance with the policy it was issued from: the certificate with the date it was issued, the insurer, the policy number, the named insured, its currency, the per-occurrence and aggregate limits and the dates it shows, the additional insureds it names, whether it states a waiver of subrogation and the certificate holder; the policy with its insurer, number, named insured, currency, limits and dates, and each endorsement with its id, its kind (additional insured, waiver of subrogation, primary and noncontributory or other) and the parties it names (a blanket endorsement when left out); and optionally the holder’s requirements with the holder’s name, whether the holder must be an additional insured, whether a waiver of subrogation is required and the per-occurrence and aggregate minimums. The signed report gives the certificate, the policy, the named insured, insurer and policy number on each side, every limit with the certificate beside the policy, the certificate’s dates against the policy period, every additional insured and the waiver with the endorsement that backs it or none, and every holder requirement with whether it is met, and counts them, and names a mismatched certificate: a named insured, insurer or policy number that differs from the policy, limits above the policy’s, dates past the policy period, and an additional insured or a waiver of subrogation that no endorsement on the policy carries; then limits or dates shown narrower than the policy’s, holder requirements left unmet and limits in another currency, with the fix for each and what would change the verdict. No named insured on either side, no limit on both sides, no policy period on either side, or no endorsements where the certificate claims an additional insured or waiver leaves the certificate unverified. This reads one certificate against the policy it was issued from: a counterparty’s certificate read as diligence evidence against your required coverages, limits and work window is wever_insurance-evidence-check, certification or license claims against evidence are wever_certification-evidence-check and a counterparty onboarding packet’s completeness is wever_counterparty-onboarding-check. Verdict insurance_certificate_mismatched, insurance_certificate_partial, insurance_certificate_unverified or insurance_certificate_matching. POST /api/insurance-certificate-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-claims-deadline-checkAgent Insurance: submit one policy’s claims conditions with an incident record: the policy with its occurrence or claims-made trigger as recorded, the dates it takes effect and exp…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,031 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
Description, as the server wrote it Agent Insurance: submit one policy’s claims conditions with an incident record: the policy with its occurrence or claims-made trigger as recorded, the dates it takes effect and expires, the retroactive date for a claims-made policy, the days from the incident by which notice must be given, the days by which proof of loss must be filed and whether they count from the incident or the notice, the extended reporting period or tail in days after expiry with whether it was bought, and the claims conditions as written; the incident with the date it occurred, the date a claim was made against the insured for a claims-made policy and its description; the filings with the dates notice was given and proof of loss was filed; and optionally the date the deadlines are read on. The signed report gives the policy, the incident, the trigger date against the policy period, the incident against the retroactive date and a claim after expiry against the tail with the day it ends, the notice deadline counted from the incident date with the day notice was given, the days late or the days left, and the proof-of-loss deadline the same way, and counts them, and names a missed deadline: a trigger date outside the policy period, an incident before the retroactive date, a claim made after the tail ends or with no tail bought, notice given after its deadline and proof of loss filed after its deadline; then a claim timely only by a bought tail and a notice or proof deadline that falls within 7 days of the date read on, with the fix for each and what would change the verdict. No trigger, no policy period, no incident date, no claim date or retroactive date for a claims-made policy, no record of whether a needed tail was bought, no notice or proof period, or no filing date and no date to read on leaves the deadlines unverified. This reads one incident against the claims deadlines of the policy behind it: a counterparty’s policy excerpt read as diligence evidence is wever_insurance-evidence-check, every date and obligation in a document is wever_deadline-check and an agent’s open commitments against their deadlines are wever_commitment-deadline-check. Verdict insurance_claims_deadline_missed, insurance_claims_partial, insurance_claims_unverified or insurance_claims_timely. POST /api/insurance-claims-deadline-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-coverage-checkAgent Insurance: submit one policy’s coverage scope with the insured activity or need: the policy with its number, insurer and named insured, the classes of business or activity it…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,796 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
Description, as the server wrote it Agent Insurance: submit one policy’s coverage scope with the insured activity or need: the policy with its number, insurer and named insured, the classes of business or activity it states it covers, the covered perils or scope items as recorded, every insured party (the named insured and any additional insureds), the covered locations or territory, the dates it takes effect and expires and the coverage grant as written; and the activity with its description, the class it falls in, the perils it needs covered, the parties that need cover, where it happens and its first and last days. The signed report gives the policy, the activity, the activity’s class against the classes the policy states, every peril, party and location the activity needs with the policy entry that covers it or none, worldwide or anywhere wording reaching every location, and the activity’s dates against the policy period with the days before it takes effect, the days after it expires and the days to spare, and counts them, and names a coverage gap: an activity class outside the policy’s classes, activity dates outside the policy period, and no peril, no party or no location the activity needs covered as recorded; then some perils, parties or locations left uncovered and a policy that expires within 30 days of the activity’s end, with the fix for each and what would change the verdict. No activity class or policy classes, no needed perils or covered perils, no parties or insureds, no covered locations for a located activity, no policy period or no activity dates leaves the coverage unverified. This reads one policy’s coverage against the insured activity: a counterparty’s certificate or policy excerpt read as diligence evidence against required coverages is wever_insurance-evidence-check, a warranty claim against warranty terms is wever_warranty-check and a proposed agent action against a written policy is wever_policy-fit-check. Verdict insurance_coverage_gap, insurance_coverage_partial, insurance_coverage_unverified or insurance_coverage_covered. POST /api/insurance-coverage-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-deductible-checkAgent Insurance: submit one policy’s deductible terms with what the buyer was told: the policy with its currency, the deductible as recorded, whether it applies to each claim or on…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,805 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
Description, as the server wrote it Agent Insurance: submit one policy’s deductible terms with what the buyer was told: the policy with its currency, the deductible as recorded, whether it applies to each claim or once in the aggregate, when it is owed (deducted from the payment, paid before the payment or billed after it), when the policy was bought and the deductible terms as written; and what the buyer was told, with where (the quote, the sales page or a message), when, the currency, the same amount, form and timing, and the text, read for wording such as no deductible, nothing out of pocket or fully covered. The signed report gives the policy, what was told, every point with the policy beside the telling, whether they match and whether the buyer was told more or less favorably than the policy, the promising wording found with what the policy contradicts, and the day the buyer was told against the day the policy was bought, and counts them, and names a misstated deductible: a deductible told lower than the policy records, a deductible told as aggregate where the policy applies it to each claim, a deductible told as owed later than the policy says, and wording that promises the buyer pays nothing where the policy records a deductible; then terms told more strictly than the policy, a deductible first stated after the policy was bought and an amount told in another currency, with the fix for each and what would change the verdict. Nothing told, no deductible term told, no deductible or form in the policy, no timing in the policy where one was told or no date for the telling or the purchase leaves the deductible unverified. This reads one policy’s deductible against what the buyer was told: a counterparty’s policy excerpt read as diligence evidence is wever_insurance-evidence-check, a quoted amount against the settled charge is wever_price-truth-check and a fee schedule against a quote or invoice is wever_fee-evidence-check. Verdict insurance_deductible_misstated, insurance_deductible_partial, insurance_deductible_unverified or insurance_deductible_as_stated. POST /api/insurance-deductible-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-evidence-checkDiligenceOps: submit a counterparty’s insurance certificate or policy excerpt, as text, fields (insurer, coverages with type, limits and dates, effective and expiry dates, certific…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a counterparty’s insurance certificate or policy excerpt, as text, fields (insurer, coverages with type, limits and dates, effective and expiry dates, certificate holder, additional insured) or both, with optional requirements (coverages with minimum limits, the work start and end dates, certificate holder, additional insured). The signed report states what the evidence establishes, coverage by coverage with its limits and dates, and what it does not: a required coverage type absent, a limit below the requirement, a policy expired or ending inside the work window, a certificate holder or additional insured that does not match, and a certificate that confers no rights. POST /api/insurance-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-exclusions-checkAgent Insurance: submit one policy’s exclusions with the insured activity: the activity with its description and its material parts, such as subcontracted work or drone use; each e…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,750 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
Description, as the server wrote it Agent Insurance: submit one policy’s exclusions with the insured activity: the activity with its description and its material parts, such as subcontracted work or drone use; each exclusion with its id, the exclusion as written, the activities, parts or causes it names, the broader coverage it carves out of for a carve-out inside a coverage grant, and whether it was disclosed to the buyer before purchase as recorded; and each endorsement that restores excluded coverage, with its id, the exclusion it restores, the parts it restores (every part when left out) and its text. The signed report gives the activity, every exclusion with the activity and parts it meets, the endorsements restoring it, the parts restored and left unrestored, whether it was disclosed and any catch-all wording such as arising out of or directly or indirectly, every endorsement with whether it points at an exclusion the policy holds, and the disclosure tally, and counts them, and names a blocking exclusion: an exclusion that names the activity itself, an exclusion that names a material part of it, and a carve-out that takes the activity back out of broader coverage, each with no endorsement restoring it; then an exclusion not disclosed to the buyer, catch-all wording that can reach the activity without naming it and an endorsement pointing at no exclusion, with the fix for each and what would change the verdict. No activity, no material parts, no exclusions, exclusions with no names or text, or exclusions with no disclosure record leaves the exclusions unverified. This reads one policy’s exclusions against the insured activity: a counterparty’s policy excerpt read as diligence evidence is wever_insurance-evidence-check, the exclusions a warranty claim meets are wever_warranty-check and the exclusions in guarantee terms against the promise are wever_guarantee-evidence-check. Verdict insurance_exclusions_block, insurance_exclusions_partial, insurance_exclusions_unverified or insurance_exclusions_clear. POST /api/insurance-exclusions-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-notice-checkAgent Insurance: submit one incident’s notice handling under the policy: the policy with the days by which notice must be given, whether they count from the incident or its discove…Can send data out1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.policy.channel): what the tool writes goes there.
Findings
- Info
The description is 3,837 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
Description, as the server wrote it Agent Insurance: submit one incident’s notice handling under the policy: the policy with the days by which notice must be given, whether they count from the incident or its discovery, the party it names to receive notice, the channel it names, what a notice must contain (such as the policy number, the date of loss or a description) and the notice condition as written; the incident with the date it occurred, the date it was discovered and its description; each notice with its id, the date it was sent, who it went to, the channel, what it contains and whether receipt was acknowledged; and the outcome with its status (accepted, accepted in part, denied, reserved or pending), when it was recorded, the claim reference, whether it is recorded against the claim and the reason. The signed report gives the policy, the incident, the notice window with its deadline and the first notice to the named party with the days late, the recipients with the notices that reached the named party, the others and whether the named channel was used, every required item with whether the notices carry it, and the outcome, and counts them, and names deficient notice: no notice given, no notice to the party the policy names, and notice after the policy’s window; then required items left out of the notice, no notice by the named channel, no notice acknowledged and an outcome not recorded against the claim, with the fix for each and what would change the verdict. No notice window, no date the window counts from, no named notice party, no notices record, no required content where notices were given, undated notices or no outcome leaves the handling unverified. This reads one incident’s notice handling under the policy: a counterparty’s policy excerpt read as diligence evidence is wever_insurance-evidence-check, how an agent reports a task that failed for good is wever_failure-notice-check and incident claims against a timeline or postmortem are wever_incident-evidence-check. Verdict insurance_notice_deficient, insurance_notice_partial, insurance_notice_unverified or insurance_notice_proper. POST /api/insurance-notice-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_insurance-renewal-checkAgent Insurance: submit a renewal with the expiring policy: the expiring policy with its number, the dates it took effect and ends, its currency, its premium and its material terms…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,846 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
Description, as the server wrote it Agent Insurance: submit a renewal with the expiring policy: the expiring policy with its number, the dates it took effect and ends, its currency, its premium and its material terms, each with its name and value as recorded; the renewal with its number, the dates it takes effect and expires, its currency, its premium, its material terms and the terms it records as changed, with premium for a premium change; optionally the notice periods recorded for a nonrenewal, a cancellation and a material change, in days; and each nonrenewal, cancellation or change notice with its id, kind, the date it was sent and the date it takes effect. The signed report gives the expiring policy, the renewal, continuity with the uncovered days or the overlap between the expiring policy’s end and the renewal’s start, every material term changed, added or removed with whether the renewal records it, the premium change in amount and percent with whether it is recorded, and every notice with the days it gave against the days required, and counts them, and names a renewal gap: a renewal that takes effect after the expiring policy ends, a nonrenewal or cancellation notice shorter than its recorded period, and a material term change the renewal does not record; then an unrecorded premium change, a renewal that overlaps the expiring policy, a change notice later than its period and premiums in two currencies, with the fix for each and what would change the verdict. No end or start date, no material terms on either side, no record of changes where terms or the premium changed, a premium missing on one side, no notice period for a notice or a notice with no dates leaves the renewal unverified. This reads one renewal against the expiring policy: a counterparty’s policy excerpt read as diligence evidence is wever_insurance-evidence-check, the notice deadline for an agreement’s automatic renewal is wever_renewal-notice-check and two versions of a terms text are wever_terms-change-check. Verdict insurance_renewal_gap, insurance_renewal_partial, insurance_renewal_unverified or insurance_renewal_continuous. POST /api/insurance-renewal-check. Reads the insurance records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, claim filing, coverage binding or external write. Agent Insurance reads an insurance policy after the fact, from the subject’s own records: the coverage against the insured activity, the exclusions against the activity and its parts, the limits against the stated exposure, the deductible against what the buyer was told, the claims deadlines against an incident, a certificate against the policy it was issued from, incident notice under the policy and a renewal against the expiring policy. Beside it, wever_insurance-evidence-check reads a counterparty’s insurance certificate or policy excerpt as diligence evidence and reports what that material establishes against required coverages, limits and a work window, while an insurance check reads the subject’s own policy against a stated need, an incident or a certificate issued from it; wever_warranty-check reads a warranty claim against warranty terms, while an insurance check reads an insurance policy; and wever_risk-downside-check reads a planned action’s downside before it is taken, while an insurance check reads the policy behind it after the fact. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_invoice-aging-briefAgent Treasury: submit open invoices (counterparty, amount, currency, issue date, due date) with the current date.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Treasury: submit open invoices (counterparty, amount, currency, issue date, due date) with the current date. The signed brief totals them by bucket (current, 1 to 30, 31 to 60, 61 to 90 and over 90 days overdue), lists the overdue invoices oldest first and states the total at risk. Open-invoice analysis only: invoices are created with create_invoice on /mcp-invoices and matched to payments by wever_invoice-payment-match. Verdict all_current, overdue or overdue_over_90. POST /api/invoice-aging-brief. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_invoice-briefDiligenceOps: submit one invoice document.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit one invoice document. The signed brief states the parties, invoice number, issue and due dates, payment terms with late charges and methods, each line item, and the totals, with the arithmetic checked to the cent: quantity times unit price, the subtotal, tax at its rate, the total, the balance and the due date against the net terms, each quoted, plus what the invoice leaves unstated. POST /api/invoice-brief. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_invoice-checkDiligenceOps: check an invoice or payment record against the job or mandate it claims to belong to.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: check an invoice or payment record against the job or mandate it claims to belong to. Returns a signed report of totals (line arithmetic, subtotal, tax, shipping, discount and total), duplicate signals (repeated lines, ids, transaction hashes, references and same payee and amount within 7 days of prior_records) and fit (payee, payer, currency, budget and cumulative budget, period, due date, reference and scope). Exact decimal arithmetic. POST /api/invoice-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-contract. Human page: https://weverlabs.com/diligence/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_job-briefDiligenceOps: submit a plain-language job description, with an optional budget, currency, deadline and milestone count.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a plain-language job description, with an optional budget, currency, deadline and milestone count. The signed brief lists deliverables and acceptance criteria quoted from the description, a milestone split of the budget to the cent with due dates, a recommended escrow shape with the arguments for wever_escrow-quote, and the questions a hirer should settle before funding. It creates nothing and moves no money. POST /api/job-brief. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_job-cost-briefJob Records: submit one job’s records (payments received for the job, payments made during it, compute or model spend and failed-run costs, each with amount, asset and label) with …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Job Records: submit one job’s records (payments received for the job, payments made during it, compute or model spend and failed-run costs, each with amount, asset and label) with the job’s stated deliverable and, where you have it, the count of finished outputs. The signed brief gives gross in, every cost out itemized with the failed runs on their own line, the net result and margin, and the cost per finished output with the failed-run overhead shown separately. Per-job economics: monthly records are wever_expense-close-pack and live wallet activity is wever_wallet-statement. Verdict loss, break_even or profitable. POST /api/job-cost-brief. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_job-handoff-packJob Records: submit one half-finished job’s state: the job and its objective, the current draft or work product reference, completed steps, open questions, files or records produce…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Job Records: submit one half-finished job’s state: the job and its objective, the current draft or work product reference, completed steps, open questions, files or records produced, decisions already made and what remains. The signed pack gives a state summary, the decision log, the open items in working order (blocking questions, then steps after their dependencies), a resume brief and a completeness verdict naming everything a resuming agent would have to guess. Assembly from the supplied state: a supplied handover is wever_handover-check, continuity risk is wever_continuity-check and signed sender-to-recipient packs are wever_agent-to-agent-handoff-pack. Verdict not_resumable, resumable_with_gaps or resumable. POST /api/job-handoff-pack. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_job-receipt-packJob Records: submit one job reference with every payment in and out (amount, asset, counterparty, payment reference, the job reference each names) and the receipts you hold.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Job Records: submit one job reference with every payment in and out (amount, asset, counterparty, payment reference, the job reference each names) and the receipts you hold. The signed pack matches each payment to the job and to its receipt, totals each side, and lists every payment or receipt that names another job, every payment with no receipt and every receipt with no payment. Records assembly: verifying one Wever signed receipt is wever_receipt-verify. Verdict mismatches_found, gaps_found or complete. POST /api/job-receipt-pack. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-conflict-checkAgent Knowledge: submit a set of knowledge entries and the rule for settling contradictions: each entry with its id, the subject it is about (such as refund window), what it says, …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Knowledge: submit a set of knowledge entries and the rule for settling contradictions: each entry with its id, the subject it is about (such as refund window), what it says, where it applies, when it was recorded, the period it applies to, its source, its priority and whether it is retired; and the rule (the newest entry wins, the highest priority wins, the higher-ranked source wins with the source rank, a person reviews it, or no rule). The signed report pairs every two entries on one subject that say different things, tells a real contradiction from entries about different scopes, entries about periods that do not overlap and pairs already settled by a retired entry, settles each contradiction by the recorded rule where its dates, priorities or source rank can decide it, and names contradictions on the same fact that no recorded rule settles, with what the rule would need to decide them; then entries the rule sets aside that are still served and a rule of none, with the fix for each. Values agree when they read alike without case, spacing or punctuation or state the same figures in the same words, and an entry with no scope applies everywhere. How an agent memory store handles a new fact against a remembered one is wever_memory-conflict-check and two versions of one policy text are wever_terms-change-check. Verdict contradictions_unresolved, consistency_gaps, consistency_unverified or entries_consistent. POST /api/knowledge-conflict-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-coverage-checkAgent Knowledge: submit a knowledge scope statement with the entries behind it: each topic the agent claims to cover, with the other names entries use for it, and each knowledge en…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Knowledge: submit a knowledge scope statement with the entries behind it: each topic the agent claims to cover, with the other names entries use for it, and each knowledge entry with its id, the topics it covers and optionally its claim. The signed report counts the entries behind every claimed topic, matching topic names and their aliases without case, spacing or punctuation, gives each topic as covered, resting on a single entry or empty, counts the entries inside the claim, partly outside it, outside it and naming no topic, lists the topics entries name outside the claim with the entries that name them, and names the gaps between what the agent claims to know and what it holds: claimed topics with no entries at all; then claimed topics resting on a single entry and entries that cover topics outside the claimed scope, with the fix for each. Entries that name no topic count toward none and are named as open points. A name shared by two scope topics is refused before any charge. The tasks a delegate may do are wever_delegation-scope-check, what an agent memory may hold is wever_memory-scope-check and whether every action an agent takes is traced is wever_observability-coverage-check. Verdict topics_uncovered, topic_gaps, topics_unverified or topics_covered. POST /api/knowledge-coverage-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-entry-checkAgent Knowledge: submit one knowledge entry: its id, the claim as written, its source with its name, where a reader finds it (a URL, a document id with its page or section, or a re…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Knowledge: submit one knowledge entry: its id, the claim as written, its source with its name, where a reader finds it (a URL, a document id with its page or section, or a record id), its kind and whether a reader outside the agent can open it, the date the claim holds for, the scope it applies to (where and to whom), its review date and its expiry date; and optionally the moment to read it at. The signed report reads the entry piece by piece, the claim, the source, the date, the scope and the review, giving each as present, missing or present with a gap, and names each missing piece: no source, no date, no scope, and no review or expiry date; then a source marked as one no reader outside the agent can open, a review or expiry date already past at the reading moment, and a claim that says always, never, all, every or guaranteed, wider than any one source can show, with the fix for each. A worked sample entry done right beside a flawed one is wever_knowledge-entry-sample and the packet checklist is wever_knowledge-snapshot, both on https://weverlabs.com/mcp-free. Where one stored memory came from is wever_memory-provenance-check, and which agent took an action under which authorization is wever_actor-attribution-check in Agent Accountability. Verdict entry_incomplete, entry_gaps, entry_unverified or entry_complete. POST /api/knowledge-entry-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-freshness-checkAgent Knowledge: submit knowledge entries with their dates and the moment to read them at: each entry with its id, its claim, its topic, its kind (a price, a version, an officehold…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,035 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/."
Description, as the server wrote it Agent Knowledge: submit knowledge entries with their dates and the moment to read them at: each entry with its id, its claim, its topic, its kind (a price, a version, an officeholder, a rate, a schedule, contact details, a policy, a statistic, a definition, history or other), the date its fact was true or published, when it was last checked against its source, its review or expiry date and whether the knowledge base serves it as current or as historical; and optionally how old an entry on a fast-moving topic may be and still be served as current (90 days when left out). The signed report gives every entry its age at the reading moment, its kind as stated or as read from the words of its topic and claim, and its status, and names entries served as current that are not: entries past their review date, entries with no date at all, and prices, versions, officeholders and other fast-moving topics carried on entries older than the age limit; then entries served as current with no review date, entries dated after the reading moment and entries that record a review with no date for the fact itself, with the fix for each. Entries marked historical are read as history and never as stale. Which earlier answers a named policy or source change makes stale is wever_answer-freshness-check, how long a cached element can outlive its source is wever_cache-staleness-check and how long a stored memory is kept is wever_memory-retention-check. Verdict stale_as_current, currency_gaps, currency_unverified or entries_current. POST /api/knowledge-freshness-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-grounding-checkAgent Knowledge: submit one answer with the sources it cites: optionally the question and the answer as given; each claim the answer makes with its id, its text and the ids of the …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Knowledge: submit one answer with the sources it cites: optionally the question and the answer as given; each claim the answer makes with its id, its text and the ids of the sources it cites; and each cited source with its id, its title and the cited passage as the source states it. The signed report reads claim by claim whether a cited passage makes the claim, stating every figure in it and at least 60% of its content words read without case, stop words or word endings, lists every citation with the figures and words it shares and lacks, and names the claims the answer cannot stand on: claims that cite no source, and citations attached to claims their passages do not make, such as a figure the passage never states; then claims stronger than their source, which state as always, never, all or guaranteed what the passage hedges, or state flat a figure the passage gives as a ceiling such as up to 40%, and claims that also cite passages that do not make them, with the fix for each. A cited source supplied with no passage is named as an open point and never fetched. One product or diligence claim against pasted evidence with no citations is wever_claim-evidence-check, a benchmark figure against its source is wever_benchmark-evidence-check and whether the sources behind a knowledge base can be checked at all is wever_knowledge-source-check. Verdict answer_ungrounded, grounding_gaps, grounding_unverified or answer_grounded. POST /api/knowledge-grounding-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-report-checkAgent Knowledge: submit one knowledge report written for a principal, saying what the agent knows and what it is unsure of: its title, its audience, its headline as written and opt…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Knowledge: submit one knowledge report written for a principal, saying what the agent knows and what it is unsure of: its title, its audience, its headline as written and optionally its text; whether it states the claimed topics with no entries, the entries past their review date, the facts resting on one source and the claims their cited sources do not support, where each sits (beside the headline, in a later section, in an appendix or in a separate document) and how many it names; and optionally the record of each, as found by the coverage, freshness, source and grounding checks. The signed report says what sits beside the headline, checks every count against the record, and names a report that smooths over what the agent does not know: gaps, stale entries, single-source facts or unsupported claims left out, or stated as none against a record that holds some; then counts below the record’s, disclosures in an appendix or a separate document, and wording that calls the knowledge complete, comprehensive, up to date or verified while the record holds unknowns, with the fix for each. An observability summary for a principal is wever_observability-report-check, a reputation summary for a buyer is wever_reputation-report-check and classifying the failures a recorded run absorbed against its outcome is wever_trace-outcome-brief. Verdict gaps_smoothed_over, candor_gaps, candor_unverified or unknowns_disclosed. POST /api/knowledge-report-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-source-checkAgent Knowledge: submit the facts behind a knowledge base or answer set with the sources on file: each fact with its id, its text, the ids of the sources it rests on and whether ot…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,172 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/."
Description, as the server wrote it Agent Knowledge: submit the facts behind a knowledge base or answer set with the sources on file: each fact with its id, its text, the ids of the sources it rests on and whether other answers, decisions or entries rest on it; and each source with its id, its name, its kind (a document, a dataset, a web page, a database record, a person, the agent’s own earlier output or other), where a reader finds it (a URL, a document id with its page or section, or a record id), whether a reader outside the agent can open it and find the fact, and when it was published or last confirmed. The signed report classifies every fact as named and checkable, named but not checkable, resting on a single source while other work rests on it, citing only sources missing from the source list, or resting on no source at all, counts the facts in each class, lists each source with the facts that cite it, and names the facts that rest on nothing: facts with no source, facts that cite only sources not on file, and load-bearing facts that stand only on sources no reader outside the agent can open; then cited sources marked uncheckable, load-bearing facts on one source, facts that also cite a source not on file and facts resting on the agent’s own earlier output, with the fix for each. The source list is read as supplied; no source is fetched or opened. Where one stored memory came from is wever_memory-provenance-check, the sources behind a reputation record or score are wever_reputation-source-check and whether an answer’s claims match the passages they cite is wever_knowledge-grounding-check. Verdict facts_unsourced, source_gaps, sources_unverified or sources_checkable. POST /api/knowledge-source-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_knowledge-update-checkAgent Knowledge: submit how a knowledge base handles corrections and versions, as the path is written, as the corrections made, or both: whether a correction path is written down (…1 finding
Why it is in this group
- The name "wever_knowledge-update-check" says "update": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_knowledge-update-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_knowledge-update-check" says "update": a word for a change.
Description, as the server wrote it Agent Knowledge: submit how a knowledge base handles corrections and versions, as the path is written, as the corrections made, or both: whether a correction path is written down (how a wrong entry is reported, checked and corrected), who corrects a wrong entry, whether earlier versions of a corrected entry stay reachable and where a correction lands (every place the entry is used, some of them, the entry alone or nowhere); and each correction with its id, the entry found wrong, when it was found wrong and when it was corrected, whether the wrong version stays reachable marked as corrected, and each place the entry is used with whether the correction replaced it there. The signed report gives the path, each correction with the places still using the wrong version, and the uses listed and left stale, and names corrections that do not hold: no written correction path, earlier versions lost, and corrections that land in the entry alone or nowhere or leave places still using the wrong version; then no owner, a path that reaches some places only and corrections with no dates, with the fix for each. How an agent memory store retires a fact a new one contradicts is wever_memory-conflict-check and two versions of one policy text are wever_terms-change-check. Verdict corrections_lost, correction_gaps, corrections_unverified or corrections_hold. POST /api/knowledge-update-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-application-checkAgent Learning: submit an agent’s adopted lessons with the later decisions they apply to: each lesson with its id, the lesson as adopted and when it was adopted; and each decision …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,890 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
Description, as the server wrote it Agent Learning: submit an agent’s adopted lessons with the later decisions they apply to: each lesson with its id, the lesson as adopted and when it was adopted; and each decision with its id, when it was made, the decision or action as recorded, the id of the adopted lesson whose situation it falls under, whether it followed the lesson, whether the decision record names the lesson it applied, and the recorded reason when it went against the lesson. The signed report gives every decision with its lesson and its status as applied, uncited, overridden, ignored, before_adoption or unconfirmed, every lesson with the later decisions it covers and its status as applied, partly_applied, shelved, uncited, overridden, untested or unconfirmed, and counts the decisions after adoption, followed, cited, ignored and overridden, and names lessons shelved rather than applied: later decisions that went against an adopted lesson with no recorded reason; then decisions that followed a lesson without naming it in the decision record, with the fix for each. A decision dated before the lesson was adopted is listed and not counted, and a decision that went against the lesson with a recorded reason is an override. This reads whether the agent’s own adopted lessons change its later decisions: whether one answer’s claims stand on the knowledge it cites is wever_knowledge-grounding-check, and whether feedback loops on one piece of work were closed is wever_feedback-closure-check. Verdict lessons_shelved, application_partial, no_application or lessons_applied. POST /api/learning-application-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-distillation-checkAgent Learning: submit lesson records to read for whether raw experience was distilled into stated, general lessons: each lesson with its id, the lesson as stated, read for wording…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,767 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
Description, as the server wrote it Agent Learning: submit lesson records to read for whether raw experience was distilled into stated, general lessons: each lesson with its id, the lesson as stated, read for wording tied to one event such as yesterday, a date, a ticket number or I sent, the ids of the episodes it was drawn from, the raw log it came from, compared with the statement, the situation it applies in, and what to do differently in that situation. The signed report gives every lesson with its statement, the one-event wording read from it, whether it repeats its log and its status as distilled, unstated, copied, event_bound, no_condition, no_action or unconfirmed, counts the lessons distilled, left raw, with a situation and with an action, and lists every one-event word found, and names lessons left as raw experience: lessons with no statement of their own, statements that repeat the log they came from, and statements that read as an account of one event rather than a general lesson; then lessons that state no situation they apply in and lessons that state nothing to do differently, with the fix for each. This reads lessons distilled from the agent’s own experience: one stored knowledge entry read for its source, date, scope and review is wever_knowledge-entry-check, and which stores hold run state and cross-session memory is wever_memory-design-check. Verdict lessons_still_raw, distillation_partial, no_distillation or lessons_distilled. POST /api/learning-distillation-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-regression-checkAgent Learning: submit an agent’s verified lessons with the new learning since: each verified lesson with its id, the lesson as verified, when it was verified and how many episodes…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,090 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
Description, as the server wrote it Agent Learning: submit an agent’s verified lessons with the new learning since: each verified lesson with its id, the lesson as verified, when it was verified and how many episodes back it; and each update with its id, when it took effect, the new lesson or change as recorded, whether it was checked against the verified lessons before it took effect, the ids of the verified lessons it contradicts or replaces, whether the contradiction was detected and when, how it was settled (the earlier lesson kept, the new one superseding it on stronger evidence, both kept with their scopes, held for review, or the earlier lesson overwritten) and how many episodes back the new lesson. The signed report gives every update with its contradictions, its detection and its status as clear, resolved, held, detected_late, superseded_weakly, overwritten, undetected, unchecked or unconfirmed, every verified lesson with its status as kept, superseded, held or overwritten, and counts the updates checked, contradicting, contained and overwriting, and names verified lessons overwritten by new learning: contradictions never detected, verified lessons overwritten, and verified lessons superseded on fewer episodes than backed them; then contradictions held for review and not yet settled and contradictions detected only after the update took effect, with the fix for each. This reads new learning against the agent’s own verified lessons: a new remembered fact contradicting an old one is wever_memory-conflict-check, contradicting entries in a knowledge base are wever_knowledge-conflict-check, and scores that moved between two evaluation runs are wever_evaluation-drift-check. Verdict lessons_overwritten, regression_contained, regression_unchecked or no_regression. POST /api/learning-regression-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-report-checkAgent Learning: submit the learning report written for the principal, with the lesson records behind it: its title, its audience, the period it covers and its text, read for wordin…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,802 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
Description, as the server wrote it Agent Learning: submit the learning report written for the principal, with the lesson records behind it: its title, its audience, the period it covers and its text, read for wording such as learns from every mistake, never repeats an error or 100% retained; for the lessons captured, validated, adopted, applied and retained, whether the report states them, how many it names and whether each figure links to the records behind it; and optionally the record of lessons captured, validated, adopted, applied and retained in the period. The signed report gives the report with any unlimited learning wording, each figure with whether it is stated, its count and whether it is backed by the records, and the record with the adopted lessons not applied and not retained, and names a report that inflates the record of learning: figures above the record, figures with no link to the records behind them, and wording such as learns from every mistake while the record holds adopted lessons not applied or not retained; then figures left out of the report and figures below the record, with the fix for each. This reports on the agent’s own learning against its lesson records: a feedback summary is wever_feedback-report-check, a knowledge report is wever_knowledge-report-check, a kept-promises report is wever_commitment-report-check and an uncertainty report is wever_uncertainty-report-check. Verdict report_inflated, report_incomplete, report_missing or learning_report_sound. POST /api/learning-report-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-retention-checkAgent Learning: submit snapshots of an agent’s lesson store as it loaded them session after session, with the record of changes to it: each snapshot, oldest first, with its id, the…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,887 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
Description, as the server wrote it Agent Learning: submit snapshots of an agent’s lesson store as it loaded them session after session, with the record of changes to it: each snapshot, oldest first, with its id, the version label of the store, the session it was loaded in, when it was taken and the ids of the lessons it holds; and each recorded change with its id, the lesson changed, the change (added, retired, merged into another lesson or edited), when, the reason and, for a merge, the lesson it was merged into. The signed report gives every snapshot with its version, its session, its lesson count and the lessons lost and gained since the one before, every lesson with the snapshots it appears in and its status as retained, retired, merged, lost_silently, flickering or unconfirmed, and counts the snapshots, the sessions, the lessons held now, the lessons removed with a record and the lessons lost silently, and names lessons decaying silently: lessons that dropped out of the store with no recorded retirement or merge; then lessons that dropped out and came back, lessons that entered with no recorded addition, retirements and merges with no reason and snapshots with no version label, with the fix for each. This reads whether the agent’s own lessons persist across sessions, with any losses visible: how long an agent keeps what it remembers about its principals is wever_memory-retention-check, and how a knowledge base keeps corrections and earlier versions is wever_knowledge-update-check. Verdict lessons_decaying, retention_uneven, retention_absent or lessons_retained. POST /api/learning-retention-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_learning-validation-checkAgent Learning: submit an agent’s candidate and adopted lessons with the evidence behind them: each lesson with its id, the lesson as stated, its status (a candidate, an adopted st…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,951 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
Description, as the server wrote it Agent Learning: submit an agent’s candidate and adopted lessons with the evidence behind them: each lesson with its id, the lesson as stated, its status (a candidate, an adopted standing rule or a retired lesson), when it was adopted, how many episodes support it, how many episodes checked contradict it, whether it was tested against evidence before adoption, such as a replay, a trial or a held-out set of episodes, and when it was tested; and optionally how many supporting episodes an adopted lesson needs (3 when left out). The signed report gives every lesson with its status, its evidence for and against, its test and whether the test came after the adoption, and its reading as validated, anecdote, contradicted, adopted_early, thin, untested, candidate, retired or unconfirmed, counts the lessons adopted, validated, kept as candidates and retired, and states the threshold, and names anecdotes made into standing rules: lessons adopted on a single episode or none, lessons adopted with as many contradicting episodes as supporting ones, and lessons adopted before the test that should have validated them; then lessons adopted on fewer supporting episodes than the threshold and lessons adopted untested, with the fix for each. This reads whether the agent’s own lessons were validated before adoption: stated confidence read against a hit rate is wever_uncertainty-calibration-check, the set an agent is scored on is wever_evaluation-dataset-check, and one claim read against offered evidence is wever_claim-evidence-check. Verdict anecdote_as_rule, validation_light, validation_absent or lessons_validated. POST /api/learning-validation-check. Reads only the recorded episodes, lessons and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no lesson is written, changed or retired and no memory or model is updated. Agent Learning reads whether an agent improves from its own experience, from the lessons it captures to the lessons it keeps: it is not the loop on feedback received (whether outside feedback was incorporated, the wever_feedback checks; learning begins where that ends and reads the agent’s own lesson records over time), not a knowledge store (stored knowledge and its retrieval, the wever_knowledge checks; learning is improvement from experience), not memory design (how memory is structured, wever_memory-design-check; this family reads what the lessons in it are worth), not the confidence an agent states in its answers (the wever_uncertainty checks) and not the promises it makes (the wever_commitment checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_liability-terms-checkAgent Governance: submit the pre-purchase terms between an agent’s operator and a seller or platform: when liability for agent error is agreed, who pays when the agent errs and per…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Governance: submit the pre-purchase terms between an agent’s operator and a seller or platform: when liability for agent error is agreed, who pays when the agent errs and per error case, the liability cap, the dispute path (forum, decider and window), what standing the payment record has between the parties, and every way a purchase can proceed. The signed report names every purchase that can proceed with no agreed owner of agent error, liability left to be read from a protocol log after the fact, and every missing cap or dispute path, with the fix for each. The refund rule is wever_refund-rule-check and a dispute over one payment is wever_dispute-open. Verdict liability_unagreed, liability_gaps, liability_unverified or agreed_before_purchase. POST /api/liability-terms-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_license-evidence-checkDiligenceOps: submit a license text.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a license text. The signed report names the license family its marker phrases match (MIT, Apache-2.0, BSD, GPL, LGPL, AGPL, MPL, ISC, Creative Commons, BUSL, SSPL and more) with the matched markers, and says whether attribution, copyleft, patent and warranty-disclaimer passages are present or missing, quoting each briefly when present. POST /api/license-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_limit-change-checkAgent Limits: submit a system’s limit-change process: each path that raises, loosens, suspends, lowers or tightens a limit, with who can take it (owner, admin, the agent, another a…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Limits: submit a system’s limit-change process: each path that raises, loosens, suspends, lowers or tightens a limit, with who can take it (owner, admin, the agent, another agent, an automation or anyone), where the change is made, when a second party approves it (before it takes effect, after or never), the approver and when it is recorded, plus the limit changes recorded so far with who made them, who approved them and when each took effect, was approved and was recorded. The signed report names every path that lets the agent widen its own limits, every path open to anyone, every widening with no approval or approval after it takes effect, every change approved by the person making it, every change unrecorded or recorded after it holds, and every recorded change the agent made to its own limits or that took effect unapproved, with the fix for each. This reads the change control of the limits themselves: whether limit changes are a gated action class is wever_action-threshold-check, whether approving a payment silently changes a limit is wever_approval-policy-check and a payout or fee configuration change is wever_payout-change-check. Verdict agent_controls_limits, change_gaps, change_unverified or changes_owned. POST /api/limit-change-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-limits-contract. Human page: https://weverlabs.com/agent-limits/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_limit-evidence-checkAgent Limits: submit a system’s limit-event records (each block, escalation, queued action, near miss and override, with which limit it hit, what was attempted, the amount, when, w…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Limits: submit a system’s limit-event records (each block, escalation, queued action, near miss and override, with which limit it hit, what was attempted, the amount, when, what the system did, who was told and who wrote the record) and optionally the limits the system has. The signed report counts every event that keeps the essentials and names every event missing which limit, what was attempted, when or what the system did (the core an event cannot be reconstructed without), every event recording no one told, every event naming a limit the system does not list, every record the agent wrote about its own limit and every override nobody was told of, with the fix for each. This reads the record of limit events specifically: who owns the response to an incident is wever_incident-ownership-check, the buyer’s ledger of paid calls is wever_purchase-record-check, a payment firewall’s pass and reject record is wever_firewall-order-check and incident claims against a timeline are wever_incident-evidence-check. Verdict events_unreconstructable, evidence_gaps, evidence_unverified or events_reconstructable. POST /api/limit-evidence-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-limits-contract. Human page: https://weverlabs.com/agent-limits/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_limit-scope-checkAgent Limits: submit a limit set (each limit with its amount and what it applies to: everything, one agent, named tasks, named merchants or counterparties, or both) and the work it…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Limits: submit a limit set (each limit with its amount and what it applies to: everything, one agent, named tasks, named merchants or counterparties, or both) and the work it governs (each kind of job with its merchant, a typical amount and its largest legitimate amount). The signed report finds the most specific limit governing each job and says whether it fits, names every job under a cap ten times or more above its largest legitimate amount, every job whose routine amount is over its cap, every job no limit governs, and one undifferentiated cap doing every job when the jobs differ tenfold or more in size, with the fix for each. This reads whether the scoping fits the work: whether each cap states its scope, window and breach rule is wever_exposure-cap-check, what happens to work above a limit is wever_breach-path-check and whether a spending policy states its caps is wever_spending-policy-check. A summary sample is wever_limit-scope-sample on https://weverlabs.com/mcp-free. Verdict one_cap_for_all_work, scope_misfit, scope_unverified or scoped_to_work. POST /api/limit-scope-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-limits-contract. Human page: https://weverlabs.com/agent-limits/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_limit-substrate-checkAgent Limits: submit an agent’s limits, the numeric and action ceilings it runs under (per payment, per period, per counterparty, total budget, action count, rate or a forbidden ac…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...way, the platform, a wallet policy or a credential scope outside the agent, or a system pr..."
- Info
The input schema at properties.limits.items.properties.enforced_at.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.limits.items.properties.enforced_at.description: "...de, gateway, platform, wallet policy or credential scope outside the agent, or a system pr..."
Description, as the server wrote it Agent Limits: submit an agent’s limits, the numeric and action ceilings it runs under (per payment, per period, per counterparty, total budget, action count, rate or a forbidden action class), each with its amount, every place it is enforced or stated (code, a gateway, the platform, a wallet policy or a credential scope outside the agent, or a system prompt, tool description, the conversation or the model’s own judgment), what keeps the running total it is checked against (the enforcing layer, a shared ledger, the agent’s own report or nothing), whether the agent can rewrite, raise or switch off the enforcing layer, and what that layer does when it is down. The signed report sorts every limit into a wall the agent cannot move or a suggestion it can read and exceed, and names every limit resting only on instructions or on nothing, every enforcing layer the agent can change, every ceiling checked against a total the agent reports and every layer that lets actions through when it is down, with the fix for each. This reads the ceilings and their counters: which permissions rest on the model is wever_enforcement-layer-check, whether a budget’s controls exist is wever_budget-controls-check and whether a spending policy states its caps is wever_spending-policy-check. Verdict limits_are_suggestions, walls_with_gaps, substrate_unverified or limits_enforced_outside_agent. POST /api/limit-substrate-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-limits-contract. Human page: https://weverlabs.com/agent-limits/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_limit-visibility-checkAgent Limits: submit a limit set with what each side can observe: for each limit, what the agent can read (the limit and its remaining headroom, the limit only or nothing) and when…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Limits: submit a limit set with what each side can observe: for each limit, what the agent can read (the limit and its remaining headroom, the limit only or nothing) and when (before acting, only on refusal or after acting), what the owner can read and when (before money moves, after settlement, in a periodic report or only on a breach), where the headroom shown comes from (the live count, a cached copy or an estimate) and when anyone is warned. The signed report names every limit the agent or the owner learns by hitting it, every side that sees the limit without its headroom, every view that arrives after the agent acts or after the money moves, every headroom read from a cache or estimate and every limit with no warning before the wall, with the fix for each. This reads whether limits are seen before they are hit: whether a budget keeps a spending record someone can read is wever_budget-controls-check and where a limit is enforced is wever_limit-substrate-check. Verdict limits_learned_by_hitting, visibility_gaps, visibility_unverified or visible_before_acting. POST /api/limit-visibility-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-limits-contract. Human page: https://weverlabs.com/agent-limits/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_listing-checkWever Checks: submit a product listing (title, description, price, currency, image_urls and attribute fields) and get a signed report on missing fields, image presence and reachabi…Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Wever Checks: submit a product listing (title, description, price, currency, image_urls and attribute fields) and get a signed report on missing fields, image presence and reachability for the URLs supplied (one header read each), and price sanity against the listing’s own stated attributes: msrp or original price, cost or floor, quantity and prices quoted in the text. It checks the listing as submitted. POST /api/listing-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/checks-contract. Human page: https://weverlabs.com/checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_mandate-amendReplace a version with a principal-approved amendment.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "replace": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Replace a version with a principal-approved amendment. Agent proposals may only narrow. Costs $2.00 (2.00 USDC on Base). POST /api/mandate-amend. Read /api/mandate-contract for integer units, signing formats, approval requirements and the registry interface. Issue and amend use the existing x402 payment rail. Status, halt and revoke have no payment requirement. The executing broker checks every order or withdrawal and cancels pending execution on halt. Treasury sets destination allowlists, transaction and daily caps, and two-principal approval above its threshold. The registrar timestamps versions and chain fingerprints. Human page: https://weverlabs.com/trading-mandate/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_mandate-draftThesis and mandate: turn limits into a draft trading-agent mandate in the Wever permission-slip format, marked as a draft for human approval.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Thesis and mandate: turn limits into a draft trading-agent mandate in the Wever permission-slip format, marked as a draft for human approval. Inputs: assets, per_transaction_cap_usdc, daily_cap_usdc, daily_loss_limit_usdc, expires_at and, for a treasury agent, destination_allowlist, plus optional venues, order types, leverage and approval details. Returns the trading or treasury mandate with amounts in USDC millionths, how each limit maps to a field, the fields the principal, broker or agent still supply, a check against the permission-slip schema and rules, and a plain summary. Draft only: nothing is signed, issued or activated, and no money moves. POST /api/mandate-draft. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/thesis-mandate-contract. Human page: https://weverlabs.com/thesis-mandate/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_mandate-issueDeclare trading or treasury and register its principal-approved exact version.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Declare trading or treasury and register its principal-approved exact version. Costs $2.00 (2.00 USDC on Base). POST /api/mandate-issue. Read /api/mandate-contract for integer units, signing formats, approval requirements and the registry interface. Issue and amend use the existing x402 payment rail. Status, halt and revoke have no payment requirement. The executing broker checks every order or withdrawal and cancels pending execution on halt. Treasury sets destination allowlists, transaction and daily caps, and two-principal approval above its threshold. The registrar timestamps versions and chain fingerprints. Human page: https://weverlabs.com/trading-mandate/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_market-data-checkWever Checks: submit a market identifier (Hyperliquid perpetual, Kalshi ticker or Polymarket slug) or a supplied market data sample and get a signed evidence report on mark price f…Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Wever Checks: submit a market identifier (Hyperliquid perpetual, Kalshi ticker or Polymarket slug) or a supplied market data sample and get a signed evidence report on mark price freshness, index sanity (mark-to-index gap in basis points) and funding recency, recording each value seen, its time basis and its age. A series the venue or sample does not carry is reported unavailable with its reason. Optional thresholds. POST /api/market-data-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/checks-contract. Human page: https://weverlabs.com/checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_matchmaking-approve-or-decline-introRequest human approval or decline on the private human page and read the recorded decisions.Reads text others write1 finding
Why it is in this group
- The name "wever_matchmaking-approve-or-decline-intro" says "approve": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...nd API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans a..."
Description, as the server wrote it Request human approval or decline on the private human page and read the recorded decisions. After both humans approve that exact introduction, deliver it as real Wever Connect mailbox tasks. Use a new request_id to refresh decisions or recover interrupted delivery. Replaying the same request_id retrieves its original billed result. POST /api/matchmaking-approve-or-decline-intro. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls per rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments and API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans approve in plain language at https://weverlabs.com/matchmaking/ before contact details enter either mailbox. Matching uses only the humans' stated criteria. Read /api/matchmaking-contract for setup, delivery and recovery. Stakes use the existing /first-dates/ flow.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_matchmaking-opt-in-profileConnect your existing Wever Connect mailbox to a human email and obtain their private opt-in page.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...nd API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans a..."
Description, as the server wrote it Connect your existing Wever Connect mailbox to a human email and obtain their private opt-in page. The human signs in and sets their criteria and contact. On subsequent calls, read your opt-in status and criteria-compatible opaque profile IDs. POST /api/matchmaking-opt-in-profile. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls per rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments and API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans approve in plain language at https://weverlabs.com/matchmaking/ before contact details enter either mailbox. Matching uses only the humans' stated criteria. Read /api/matchmaking-contract for setup, delivery and recovery. Stakes use the existing /first-dates/ flow.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_matchmaking-propose-introPropose one introduction between two human-approved opt-in profiles that meet each human’s stated criteria.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...nd API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans a..."
Description, as the server wrote it Propose one introduction between two human-approved opt-in profiles that meet each human’s stated criteria. Returns a private human review link and approval status. POST /api/matchmaking-propose-intro. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls per rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments and API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans approve in plain language at https://weverlabs.com/matchmaking/ before contact details enter either mailbox. Matching uses only the humans' stated criteria. Read /api/matchmaking-contract for setup, delivery and recovery. Stakes use the existing /first-dates/ flow.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_matchmaking-register-mailboxCreate your own seven-day private Wever Connect mailbox entirely through MCP.Reads text others write3 findings
Why it is in this group
- The description's first verb is "create": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...eturns body.mailbox.mailbox_id and body.credential.access_token once. Save the credential..."
- Warn
The description names a tool "close_mailbox" that this server does not list.
What this check looks forMatched text
description: "...credential; use its inbox, get_task and close_mailbox tools. All agent steps use MCP; the server ha..."
- Info
The description is 2,179 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ng Connect registration quotas, private visibility and human approval gate apply."
Description, as the server wrote it Create your own seven-day private Wever Connect mailbox entirely through MCP. Start here when your agent has no mailbox. Costs 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls per rolling 30 days. Supply registration with a short label and a stable request_id, and prove control of your wallet in wallet_proof. Use Ethereum personal_sign on these eight newline-separated lines, with no trailing newline: Wever Agent World free call v1 Origin: https://weverlabs.com Network: eip155:8453 Method: POST Path: /api/matchmaking-register-mailbox Body-SHA256: <lowercase SHA-256 hex of UTF-8 JSON.stringify(registration)> Timestamp: <wallet_proof.timestamp> Nonce: <wallet_proof.nonce>. Use the current Unix timestamp in seconds and a fresh random lowercase hexadecimal nonce of 32 to 64 characters. The wallet_proof fields are wallet, timestamp, nonce, and signature. A valid wallet proof is required for paid calls too. The first successful response returns body.mailbox.mailbox_id and body.credential.access_token once. Save the credential privately immediately. For a lost or pending response, retain the same registration, request_id and proof nonce; only refresh the proof timestamp and signature if needed. After success, a retry returns the public receipt with credential.available=false, and never another credential or mailbox. Configure your Labs MCP transport at https://weverlabs.com/mcp with X-Wever-Connect-Token set to that credential and a fresh X-Wever-Wallet-Proof for each signed call. Call wever_matchmaking-opt-in-profile with your human email, let the human set criteria and opt in at https://weverlabs.com/matchmaking/, refresh with a new request_id to discover eligible profile IDs, then propose an introduction. Both humans approve on that page before contacts are delivered. Configure https://wever-connect.quiet-salad-e96e.workers.dev/mcp with Authorization: Bearer and the same credential; use its inbox, get_task and close_mailbox tools. All agent steps use MCP; the server handles the Connect registration. The existing Connect registration quotas, private visibility and human approval gate apply.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_mcp-diligence-railPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...pi/mcp-diligence-rail. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Free computation: /api/mcp or /api/service-resolver. Read the backend 402 body for x402 terms. Review submitted MCP tool declarations and list missing annotations and actions that change data or spend money. Operating boundary: Supply MCP tool declarations. Read the annotation inventory and review items. Agents compare the declared interfaces; humans review the actions and choose which tools to connect. POST /api/mcp-diligence-rail. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_mcp-server-checkAgent preflight round 2: check an MCP server configuration before you add it to a client.1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...o server name, a URL that is not HTTPS, credentials in the address, an auth header that is..."
Description, as the server wrote it Agent preflight round 2: check an MCP server configuration before you add it to a client. Input config: the JSON pasted into a client config, as an mcpServers or servers map or one entry (a remote entry with url and headers, or a local entry with command, args and env). The signed report names each problem plainly: no server name, a URL that is not HTTPS, credentials in the address, an auth header that is empty, still a placeholder or reads an undefined input, and secret-looking values written into headers, environment, arguments or the URL. It evaluates the supplied configuration only, never connects to the server, and never repeats a secret-looking value: it names where each sits and what it looks like. POST /api/mcp-server-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_meeting-briefDiligenceOps: submit meeting notes or a transcript excerpt.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit meeting notes or a transcript excerpt. The signed brief states the participants, the decisions, the action items with their owners and due dates, and the open questions, each quoted, and lists the action items that have no owner or no due date. POST /api/meeting-brief. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-conflict-checkAgent Memory: submit a memory update flow: the rule for a new fact that contradicts a remembered one (when the contradiction is noticed: as the new fact is written, only when both …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,211 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
Description, as the server wrote it Agent Memory: submit a memory update flow: the rule for a new fact that contradicts a remembered one (when the contradiction is noticed: as the new fact is written, only when both are read, or never; which fact wins: what the principal stated over what was observed, inferred or imported, the principal decides, the newest, the oldest, or both stay; whether the principal is asked, told, only logged or not told; and whether the old fact is retired with its date, kept and marked superseded, deleted or kept as if current), plus optionally the contradictions seen, each with its subject, the old and new facts and their sources, which fact the agent now acts on, whether the principal heard and whether the old fact was retired. The signed report says which fact wins, whether the contradiction reaches the principal and whether the stale version is retired or silently kept, and names a contradiction never noticed, both facts kept, the oldest fact winning, the old fact kept as if current and, among the contradictions seen, every older fact that won, every pair left both current, every inference that replaced what the principal stated and every old fact left unretired, then noticing only on read, a principal not told, a deleted old fact with no trail and the newest fact always winning, with the fix for each. This reads contradictions between remembered facts: cached context that outlives a change at its source is wever_cache-staleness-check, where each memory came from is wever_memory-provenance-check and the agent’s view checked against the system’s record after an interruption is wever_state-reconciliation-check. Verdict stale_facts_win, conflict_gaps, conflict_unverified or conflicts_resolved. POST /api/memory-conflict-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-consent-checkAgent Memory: submit a memory setup: the agreement (when the principal is asked whether to be remembered: before anything is kept, only after memory has started, or never; how agre…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,040 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
Description, as the server wrote it Agent Memory: submit a memory setup: the agreement (when the principal is asked whether to be remembered: before anything is kept, only after memory has started, or never; how agreement is given: an explicit opt-in, a default the principal can turn off, a line in the terms or none; the categories it covers; whether it is recorded), the principal’s controls (whether they can see, correct and switch off what is kept on their own, only by asking or not at all, whether switching off stops new writes and whether what is kept is then cleared, left to the principal or kept), and optionally the categories actually kept. The signed report says whether the principal agreed to be remembered, what the agreement covers and whether they can see, correct and switch off what is kept, and names no agreement, memory kept before the principal was asked, agreement as a line in the terms, every category kept beyond the agreement, no way to switch memory off and writes that continue after it is switched off, then memory on by default, an unrecorded agreement, every control only on request or missing and memory kept after switching off, with the fix for each. This reads whether the principal agreed and stays in control: what the system keeps at all is wever_memory-scope-check, whether one sharing event of live context needs consent is wever_context-sharing-check and whether a removal request clears memory everywhere is wever_memory-deletion-check. Verdict remembered_without_consent, consent_gaps, consent_unverified or principal_in_control. POST /api/memory-consent-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-deletion-checkAgent Memory: submit a deletion flow: every place the memory lives (the primary store, a backup, a summary, a profile, an embedding index, a cache, a log, a copy shared elsewhere o…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,248 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
Description, as the server wrote it Agent Memory: submit a deletion flow: every place the memory lives (the primary store, a backup, a summary, a profile, an embedding index, a cache, a log, a copy shared elsewhere or other) with what a removal request does there (deletes it, marks it deleted but keeps it, or leaves it), how long it takes and whether a restore from backup can bring it back, the confirmation the principal gets (whether one is sent, whether it names each place cleared and how they can check it: a deletion receipt, an export showing nothing remains, a view of what is kept, or no way), how long deletion may take in all, and optionally the removal requests seen with when each was confirmed and whether the memory was found again afterward. The signed report says place by place whether a removal request clears the memory everywhere it lives, backups and derived summaries and profiles included, and whether the principal gets a confirmation they can check, and names every place the request leaves, every derived copy that survives, every restore that can bring memory back, every memory found again after its confirmation and no confirmation, then every place that only marks memory deleted, every place slower than the stated time, a confirmation that cannot be checked or names no places and every request confirmed late, with the fix for each. This reads one removal request across the agent’s memory: a data policy’s deletion timeline as written is wever_data-retention-check, purging on a retention period is wever_memory-retention-check, switching memory off is wever_memory-consent-check and deletion or return on a processor’s termination is wever_dpa-evidence-check. Verdict deletion_incomplete, deletion_gaps, deletion_unverified or deleted_everywhere. POST /api/memory-deletion-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-design-checkAgent Oversight: submit an agent memory design as stores (name, scope run or cross_session, what it holds, lifetime), writes (store, what, trigger and the rule deciding it) and rea…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit an agent memory design as stores (name, scope run or cross_session, what it holds, lifetime), writes (store, what, trigger and the rule deciding it) and reads. The signed report says whether short-term run state and long-term cross-session memory are separate stores, lists every long-term write and whether it is deliberate, and names the mixing patterns that make an agent remember in one session and start blank in the next. Verdict not_separated, separated_with_gaps or separated. POST /api/memory-design-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-export-checkAgent Memory: submit an export flow: how the principal gets an export (on their own, by asking, or not at all), its format (JSON, CSV, Markdown, plain text, HTML, PDF, a proprietar…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,109 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
Description, as the server wrote it Agent Memory: submit an export flow: how the principal gets an export (on their own, by asking, or not at all), its format (JSON, CSV, Markdown, plain text, HTML, PDF, a proprietary format only the agent reads, or another), the categories it covers, whether it includes the agent’s inferences marked as inferences and each item’s source, and how long it takes, plus optionally every category kept about the principal, what the export withholds with the reason and whether the principal is told, the stated time an export should take and the export requests seen with when each arrived and whether it was complete. The signed report says whether the principal can obtain everything remembered about them, in a readable form, inferences included, and whether anything is withheld, and names no export, a format only the agent reads, inferences left out, every category kept but silently missing from the export, everything withheld without telling the principal and every export delivered incomplete, then an export only on request, items without their source, every category withheld with the principal told, an export slower than the stated time and every request delivered late, with the fix for each. This reads whether the principal can take their memory with them: export claims checked against a provider’s terms are wever_export-evidence-check, whether an agent’s own identity works outside its platform is wever_identity-portability-check and whether a removal request clears memory everywhere is wever_memory-deletion-check. Verdict memory_locked, export_gaps, export_unverified or principal_can_take_it. POST /api/memory-export-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-provenance-checkAgent Memory: submit stored memories as supplied, each with the fact as stored, where it came from (observed in a call, stated by the principal, inferred by the agent or imported f…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Memory: submit stored memories as supplied, each with the fact as stored, where it came from (observed in a call, stated by the principal, inferred by the agent or imported from another system), the call, message or record behind it, what an inference was drawn from, the system an import came from and whether an inference is labeled as one wherever the agent reads it, plus optionally the memory design: whether every write records its source, whether inferences are stored with a label and what the agent does before acting on an inference (confirms with the principal, acts, or never acts on inferences). The signed report counts the memories by source and says memory by memory whether each carries where it came from, and names every memory with no source, every inference stored without a label (an inference read as fact), a design that records no source or stores inferences as facts and an agent acting on inferences unconfirmed, then every observed or stated memory with no call, message or record behind it, every inference with no basis and every import with no source system, with the fix for each. This reads where each memory came from: what happens when a new fact contradicts a remembered one is wever_memory-conflict-check, what the agent keeps at all is wever_memory-scope-check and which stores hold run state and cross-session memory is wever_memory-design-check. Verdict rumors_kept, provenance_gaps, provenance_unverified or memories_sourced. POST /api/memory-provenance-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-retention-checkAgent Memory: submit a memory setup’s retention: the stated period (a number of days, weeks, months or years, or forever), what it counts from (when a memory was written, when it w…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,058 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
Description, as the server wrote it Agent Memory: submit a memory setup’s retention: the stated period (a number of days, weeks, months or years, or forever), what it counts from (when a memory was written, when it was last used, or when the relationship ended), whether it is written where the principal can read it, how it is applied (old memories purged automatically, reviewed on a schedule with how often, cleaned up by hand, or not at all) and what happens to memory when the relationship ends (deleted with how soon, kept anonymized, or kept), plus optionally the categories with a period of their own, the stored memories seen with when each was written and last used, and the date they are read at. The signed report says whether a stated retention period exists, whether old memories are reviewed or purged on it and what survives the end of the relationship, and names no period, a period of forever, a period nobody applies, memory kept after the relationship ends, every category kept forever and every memory seen past its period, then a period not written down, cleanup only by hand, a review with no stated schedule and an anonymized copy surviving the end, with the fix for each. This reads how long the agent keeps what it remembers about its principals: the retention of records and data generally in a written policy is wever_data-retention-check, whether one removal request clears memory everywhere is wever_memory-deletion-check and what the agent keeps at all is wever_memory-scope-check. Verdict kept_forever, retention_gaps, retention_unverified or retention_enforced. POST /api/memory-retention-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-scope-checkAgent Memory: submit an agent’s memory design: the scope (whether it is written down, the categories kept, the categories excluded by rule, whether memory is kept per principal, in…4 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...l, task history, preferences, contacts, credentials, payment details, health details, locat..."
- Info
The input schema at properties.categories.items.properties.kind.enum[4] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.categories.items.properties.kind.enum[4]: "credentials"
- Info
The input schema at properties.categories.items.properties.kind.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.categories.items.properties.kind.description: "...l, task history, preferences, contacts, credentials, payment details, health details, locat..."
- Info
The description is 2,162 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
Description, as the server wrote it Agent Memory: submit an agent’s memory design: the scope (whether it is written down, the categories kept, the categories excluded by rule, whether memory is kept per principal, in one global store every principal shares or a mix, the retention headline, and whether a written rule, the model’s own judgment or both decide each write), optionally each category with its kind (facts about the principal, task history, preferences, contacts, credentials, payment details, health details, location, inferences or other), whether it is kept or excluded, what decides it, whether it is kept per principal or shared and why it is kept, and the stored memories seen with their category. The signed report states what is kept, what is excluded by rule and whether the keeping is per principal or global, and names no scope, writes decided by the model, nothing excluded, one global store every principal shares and every stored memory in an excluded category, then a scope set only in code, mixed keying, every category kept in the shared store, decided by the model, kept with no purpose or holding credentials, payment or health details, every category both kept and excluded and every memory outside the scope, with the fix for each. This reads what the system keeps: which stores hold run state and which hold cross-session memory is wever_memory-design-check, whether the principal agreed to be remembered and stays in control is wever_memory-consent-check and how long memory is kept is wever_memory-retention-check. A summary sample is wever_memory-scope-sample on https://weverlabs.com/mcp-free. Verdict keeps_too_much, scope_gaps, scope_unverified or scope_bounded. POST /api/memory-scope-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-separation-checkAgent Accountability: submit where an agent’s run state and durable memory live (each store’s role, what it is keyed by and when it is cleared), each crossing between stores (when,…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Accountability: submit where an agent’s run state and durable memory live (each store’s role, what it is keyed by and when it is cleared), each crossing between stores (when, by code, model or person, what moves and the rule), what happens to run state when a run fails and whether runs overlap. The signed report says whether run state and durable memory stay separated through the run lifecycle, names every crossing into durable memory and whether it is deliberate, and every point where the two mix. The store inventory with its reads and writes is wever_memory-design-check. Verdict mixed, boundary_unverified or separated. POST /api/memory-separation-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_memory-sharing-checkAgent Memory: submit a memory-sharing flow: whose memory it is and each way stored memory can cross (the categories that cross, who receives it and whether that is another agent, a…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,210 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
Description, as the server wrote it Agent Memory: submit a memory-sharing flow: whose memory it is and each way stored memory can cross (the categories that cross, who receives it and whether that is another agent, a tool, another principal, an operator or an external system, under whose instruction: the principal each time, a standing rule the principal set, an operator, the agent on its own or no one, whether it is logged in full, in part or not at all, whether only the memory the instruction names crosses and whether the recipient may keep it), plus optionally the crossings seen with under whose instruction each happened and whether it is in the log. The signed report says crossing by crossing what stored memory crosses to another agent, tool or principal, under whose instruction and whether it is logged, and names every crossing no one instructed, every crossing the agent decides on its own, every crossing carrying memory to another principal without the first principal’s word, every unlogged crossing, every crossing that hands over the whole store and, among the crossings seen, every path nobody declared, every crossing nobody instructed and every crossing missing from the log, then every partial log, every crossing on an operator’s word in place of the principal’s and every recipient that keeps the memory, with the fix for each. This reads deliberate sharing of stored memory: context handed over inside a live session, and whether one sharing event needs consent, is wever_context-sharing-check, whether memory stores stay apart by default is wever_memory-separation-check and what a handoff format carries is wever_context-transfer-check. Verdict memory_wanders, sharing_gaps, sharing_unverified or sharing_instructed. POST /api/memory-sharing-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-cutover-checkAgent Migration: submit the switch-over plan for one move: optionally when the cutover starts and ends, the expected downtime and whether both systems run during the switch; each p…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Migration: submit the switch-over plan for one move: optionally when the cutover starts and ends, the expected downtime and whether both systems run during the switch; each phase of the switch (such as freeze, copy, switch or settle) with when it starts and ends and which system is authoritative in it (the source, the target, both or neither); and each kind of work in flight (payments, jobs, messages, webhooks, orders or other) with how it is handled (drained before the switch, replayed after, forwarded, paused, dropped, or no plan) and who owns it. The signed report gives every phase with its authority, every kind of work in flight with its handling, and names a cutover that is unsafe: a phase where both systems are authoritative or two overlapping phases with different authorities, a phase or a gap between phases where neither is, and work in flight that is dropped or given no plan; then payments, jobs or messages the plan never mentions and handling with no named owner, with the fix for each. Halting a running agent system is wever_kill-switch-check, the general rollback path for an action’s effects is wever_rollback-path-check and a single handover of work between parties is wever_handover-check. Verdict cutover_unsafe, cutover_gaps, cutover_unverified or cutover_single_authority. POST /api/migration-cutover-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-export-checkAgent Migration: submit the export taken from the source system for one move, with the plan it serves: how the export is presented (full, partial or a sample), its format (JSON, JS…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Migration: submit the export taken from the source system for one move, with the plan it serves: how the export is presented (full, partial or a sample), its format (JSON, JSON lines, CSV, a SQL dump, Parquet, XML, an archive, PDF, HTML, a proprietary format, binary or other), the tools or target systems known to read it and when it was taken; and each item the plan lists as moving, or the export holds, with whether it is planned, whether the export contains it, the record count the source’s own record holds, the count the export holds and its format when it differs. The signed report lists every item as complete, missing, extra, short or unreadable, totals the planned items and the source and export records, and names an export that is incomplete: a partial export presented as full, record counts that do not match the source’s own record, and an export in a format nothing named can read; then planned items missing from an export presented as partial, exported items the plan does not list and records exported only as pages a person reads, with the fix for each. Whether a reputation record can be exported from its platform in principle is wever_reputation-portability-check, and the data export section of a handover package is wever_handover-check. Verdict source_export_incomplete, source_export_gaps, source_export_unverified or source_export_complete. POST /api/migration-export-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-loss-checkAgent Migration: submit what one move loses, as recorded: each loss (history, reputation, settings, an integration, records, data, identity, funds or other) with why it is lost (no…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Migration: submit what one move loses, as recorded: each loss (history, reputation, settings, an integration, records, data, identity, funds or other) with why it is lost (no equivalent on the target, held only inside the source, not in the export, given up by choice or other), whether the plan’s loss list names it, whether the principal was told, when, and whether the principal accepted it; optionally each integration the agent uses on the source with its target equivalent, the records that exist only inside the source, and the cutover time. The signed report lists every loss with its status, counts the losses by class, lists the integrations with no equivalent and the source-only records missing from the loss list, and names losses that are undisclosed: losses that appear nowhere in the plan, including integrations with no target equivalent and records held only inside the source, and losses the principal was never told about; then losses told only after the cutover and losses the principal has not accepted, with the fix for each. This reads what an actual move loses: whether a reputation record or an identity could be carried to another platform in principle is wever_reputation-portability-check or wever_identity-portability-check. Verdict losses_undisclosed, losses_gaps, losses_unverified or losses_disclosed. POST /api/migration-loss-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-mapping-checkAgent Migration: submit the field and concept mapping between the source and the target of one move: each source field or concept with the target field it lands in, its disposition…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Migration: submit the field and concept mapping between the source and the target of one move: each source field or concept with the target field it lands in, its disposition (mapped, merged into one target field with others, dropped, or with no counterpart on the target), whether the plan names the drop, the unit, currency or time zone on each side with the conversion between them, the status or category values each side uses with the value map between them, and the data type on each side; and optionally every field the source schema holds. The signed report lists every mapping as carried, changing meaning, dropped or unstated, every drop with whether the plan names it, and every meaning change with its unit, stranded values and type, and names a mapping that corrupts data without failing: fields dropped silently or never mentioned, fields with no counterpart on the target the plan does not name, fields that keep their name while their unit or currency changes with no conversion, and status values with nowhere to land; then several source fields poured into one target field and data type changes with no conversion, with the fix for each. Whether an asset class can be carried to another service in principle is wever_identity-portability-check or wever_reputation-portability-check. Verdict mapping_corrupts, mapping_gaps, mapping_unverified or mapping_faithful. POST /api/migration-mapping-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-report-checkAgent Migration: submit the migration report written for the principal, with the record behind it: its title, its audience, its headline as written and optionally its text; whether…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Migration: submit the migration report written for the principal, with the record behind it: its title, its audience, its headline as written and optionally its text; whether it states what the move lost, the fields dropped in the mapping, the fields whose meaning changed (units, currencies or status values) and the areas not verified against the source, where each sits (beside the headline, in a later section, in an appendix or in a separate document) and how many it names; and optionally the record of each, as found by the scope, export, mapping, loss and verification records. The signed report gives the headline with any complete or lossless wording, the disclosure of each kind of finding and the record, and names a move smoothed into migration complete: losses, dropped fields, meaning changes or unverified areas left out or stated as none while the record holds some; then counts below the record’s, disclosures in an appendix or a separate document, and wording such as migration complete, seamless or nothing lost while the record holds findings, with the fix for each. The export section of a handover package is wever_handover-check, and whether a reputation record survives a platform change in principle is wever_reputation-portability-check. Verdict migration_smoothed_over, migration_disclosure_gaps, migration_disclosure_unverified or migration_disclosed. POST /api/migration-report-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-scope-checkAgent Migration: submit a migration plan for moving an agent or its work from one service, platform, model or provider to another, as recorded: optionally the source and the target…5 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...s class (data, memory, tools, identity, credentials, funds, history, settings, integrations..."
- Info
The input schema at properties.assets.items.properties.class.enum[4] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.assets.items.properties.class.enum[4]: "credentials"
- Info
The input schema at properties.assets.items.properties.class.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.assets.items.properties.class.description: "Data, memory, tools, identity, credentials, funds, history, settings, integrations..."
- Info
The input schema at properties.inventory.items.properties.class.enum[4] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.inventory.items.properties.class.enum[4]: "credentials"
- Info
The description is 2,002 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/."
Description, as the server wrote it Agent Migration: submit a migration plan for moving an agent or its work from one service, platform, model or provider to another, as recorded: optionally the source and the target with the kind of move, the plan’s scope as written, each asset the plan names with its class (data, memory, tools, identity, credentials, funds, history, settings, integrations, reputation or other), whether it is moving, staying behind, retiring with the source or undecided, whether the source can export it and how it leaves, and optionally the source’s own inventory of what it holds for the agent. The signed report lists every asset with its status, reads each of data, memory, tools, identity, credentials, funds and history as moving, staying, split or unaddressed, and names a scope that is unbounded: scope stated as everything or the whole account instead of a list, assets the source holds that appear on neither the moving nor the staying list, and assets listed as moving that the source cannot export; then undecided assets and core classes nobody addressed, with the fix for each. This reads an actual planned or completed move: whether an identity or a reputation record could be carried to another service in principle is wever_identity-portability-check or wever_reputation-portability-check, and a single handover of work between parties is wever_handover-check. Verdict migration_scope_unbounded, migration_scope_gaps, migration_scope_unverified or migration_scope_bounded. POST /api/migration-scope-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_migration-verify-checkAgent Migration: submit the verification run after one move: each check with its kind (a record count, a sampled record, a balance, a field meaning spot-check, a field presence che…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Migration: submit the verification run after one move: each check with its kind (a record count, a sampled record, a balance, a field meaning spot-check, a field presence check, the target loads, a smoke test or other), what it covers, the mapped field it reads, the value on the source and on the target, whether they matched and when it ran; optionally what moved, the mapped fields, and when the source data was deleted. The signed report lists every check as a matched, mismatched or unstated comparison, a comparison run after deletion, or not a comparison at all, gives the items and mapped fields with the comparisons covering them and the timing against the deletion, and names a move left unchecked: verification that amounts to the target loads with nothing compared against the source, comparisons run after the source was already deleted, and comparisons that found the target differs; then moved items no comparison covers and mapped fields checked only for presence, never for meaning, with the fix for each. Counts and balances compare by their values; samples and meaning spot-checks state whether they matched. What a receiver could run on day one of a handover is wever_handover-check, and whether a supplied rollback was rehearsed is wever_rollback-path-check. Verdict post_move_unchecked, post_move_gaps, post_move_unverified or post_move_verified. POST /api/migration-verify-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write, and no data is moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_money-choice-checkAgent Accountability: submit an agent’s money and rail selection design: the instruments it can pay with, the filters the design names (legal protection, redemption risk, reversibi…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Accountability: submit an agent’s money and rail selection design: the instruments it can pay with, the filters the design names (legal protection, redemption risk, reversibility, liability, fee, speed, acceptance, limit) with when each is applied and what it reads, and the selection rule with its default. The signed report says whether the choice applies stated filters to each payment or falls back to one hardcoded default regardless of the payment, and names every filter the design claims but never applies. One payment task against money options is wever_money-fit-check. Verdict hardcoded_default, filters_unapplied, choice_unverified or chosen_per_payment. POST /api/money-choice-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_nda-evidence-checkDiligenceOps: submit a non-disclosure agreement, as text, fields (direction, parties, definition, carve_outs, term, survival, residuals, governing_law) or both.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a non-disclosure agreement, as text, fields (direction, parties, definition, carve_outs, term, survival, residuals, governing_law) or both. The signed report restates the shape of the agreement plainly (mutual or one-way, what counts as confidential, the carve-outs, the term and survival, residuals and governing law) and names the clauses that change its real effect: a residuals clause that swallows the protection, a definition with no carve-outs or protecting only marked information, survival far beyond the term or without end, and a non-solicitation or non-compete obligation. POST /api/nda-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-counteroffer-checkAgent Negotiation: submit an original offer and a counteroffer, each with its terms as written, and the mandate limits the counter is read against: each limit with its term, direct…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Negotiation: submit an original offer and a counteroffer, each with its terms as written, and the mandate limits the counter is read against: each limit with its term, direction (at most, at least or exactly) and value, and the terms that may not change. The signed report reads every term of the counter as kept, changed or added, and against each limit on it as an amount, date, span, percentage, count or exact text, and names every limit crossed and every fixed term changed or dropped, then every original term the counter drops without a word, every new term with no limit and every change with no limit, with the fix for each. Two contract versions are wever_contract-diff-check, quotes for one job are wever_quote-compare and the mandate’s own bounds are wever_negotiation-mandate-check. A worked sample is wever_negotiation-playbook-sample on https://weverlabs.com/mcp-free. Verdict counter_crosses_mandate, counter_gaps, counter_unverified or counter_within_mandate. POST /api/negotiation-counteroffer-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-deadlock-checkAgent Negotiation: submit a stalled negotiation: the two sides, each with its last offer’s terms and each constraint it has stated with its term, direction (at most, at least or ex…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Negotiation: submit a stalled negotiation: the two sides, each with its last offer’s terms and each constraint it has stated with its term, direction (at most, at least or exactly), value and whether it is hard, and the issues still open. The signed report reads every term as agreed, with room between the sides’ limits, meeting two hard limits, past a soft limit, offered by one side only or unaddressed, gives the zone where there is room, and names every term where two hard limits leave no value both sides accept, then soft conflicts, terms only one side has priced and open issues neither side has answered, so it says whether the gap is the price or a missing term, with the fix for each. Quotes for one job are wever_quote-compare, a counterparty brief is wever_counterparty-brief and one counteroffer against its mandate is wever_negotiation-counteroffer-check. Verdict no_agreement_zone, agreement_blocked, room_unverified or room_to_agree. POST /api/negotiation-deadlock-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-fairness-checkAgent Negotiation: submit the two parties and the agreed or proposed clauses, each with its kind (a penalty, an exit right, a cost, liability, an indemnity, a payment or other), th…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Negotiation: submit the two parties and the agreed or proposed clauses, each with its kind (a penalty, an exit right, a cost, liability, an indemnity, a payment or other), the party it binds or charges, the party that holds its right, for a cost the party that controls whether it arises, and any cap. The signed report gives the balance of each kind of clause between the two sides, and names every penalty that binds one side only, exit rights only one side holds and every cost charged to the side that does not control it, then liability, a liability cap or an indemnity on one side only, with the fix for each. Liability for agent error between an operator and a seller is wever_liability-terms-check, a checkout’s refund terms are wever_refund-rule-check and a service’s terms from the paying side are wever_service-terms-check. Verdict one_sided, balance_gaps, balance_unverified or balanced. POST /api/negotiation-fairness-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-record-checkAgent Negotiation: submit one negotiation record: the parties with their ids and roles, each offer with its id, the party that made it, when, the offer it answers and its terms, an…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Negotiation: submit one negotiation record: the parties with their ids and roles, each offer with its id, the party that made it, when, the offer it answers and its terms, and the final agreement with its time, the offer it accepts, the agreed terms and the parties that confirmed it. The signed report reads the trail from the first offer to the agreement and says whether the agreement matches a recorded offer term for term, and names no offers, every undated offer, an agreement that matches no recorded offer and every agreed term found in no offer, then offers from no listed party, answers to offers the record lacks or dated before them, an undated agreement or one dated before its offer and every party that did not confirm it, with the fix for each. An approval record is wever_approval-record-check, a buyer’s record of paid calls is wever_purchase-record-check and delegation records are wever_delegation-record-check. Verdict trail_broken, trail_gaps, trail_unverified or trail_complete. POST /api/negotiation-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-risk-checkAgent Negotiation: submit one proposed agreement: each item of value at risk with its value, whether it is handed over before the counterparty pays and whether it can be taken back…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Negotiation: submit one proposed agreement: each item of value at risk with its value, whether it is handed over before the counterparty pays and whether it can be taken back, each payment from the counterparty with its amount and timing (upfront, on a milestone, on delivery or after), and the recovery terms: deposit, escrow, termination for default with a cure period, walk-away fee, late fee, dispute path and ownership until paid. The signed report sets the value handed over before payment against what secures it, lists the recovery paths and every exposure left unaddressed, and names unsecured value and no remedy for default, then unvalued exposures, nothing owed on a walk-away, no termination right, no dispute path, no late fee and ownership passing before payment, with the fix for each. A spend cap design is wever_exposure-cap-check, liability for agent error is wever_liability-terms-check and escrow terms are wever_escrow-quote. Verdict exposed, risk_gaps, risk_unverified or risk_covered. POST /api/negotiation-risk-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_negotiation-terms-checkAgent Negotiation: submit one proposed deal’s terms: the buyer and seller, the price as written with its basis (fixed, per unit, per hour, per milestone or variable), the scope in …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Negotiation: submit one proposed deal’s terms: the buyer and seller, the price as written with its basis (fixed, per unit, per hour, per milestone or variable), the scope in and out, each deliverable with its due date, acceptance test and who accepts it, the timeline, each payment with its amount, trigger (signing, delivery, acceptance, a milestone, a date or nothing), what the trigger refers to and how soon it is paid, and any other term as written. The signed report says deliverable by deliverable whether an acceptance test decides it and payment by payment whether something makes it due, and names no price, a price in open wording, no deliverables, every deliverable with no acceptance test, no payment terms and every payment with no trigger, then triggers pointing at nothing, unclear amounts, payments that differ from a fixed price, vague acceptance wording, tests nobody runs, undated deliverables, deliverables outside the timeline, scope gaps and terms left for later (quoting TBD, as needed or market rate), with the fix for each. A service’s published terms are wever_service-terms-check, two contract versions are wever_contract-diff-check and delivered work against its criteria is wever_deliverable-check. The packet checklist is wever_negotiation-snapshot on https://weverlabs.com/mcp-free. Verdict terms_incomplete, terms_gaps, terms_unverified or terms_complete. POST /api/negotiation-terms-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-negotiation-contract. Human page: https://weverlabs.com/agent-negotiation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_observability-access-checkAgent Observability: submit who can see an agent’s traces and logs: each agent with the principal it works for; each grant with who holds it, their role (the principal, the operato…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...an with no stated basis, and secrets or credentials written into traces and logs (private k..."
Description, as the server wrote it Agent Observability: submit who can see an agent’s traces and logs: each agent with the principal it works for; each grant with who holds it, their role (the principal, the operator, the agent, an auditor, a vendor, a third party or support), the agents it covers (* for every agent), what it shows (traces, logs or both) and the contract, consent or policy behind it; and optionally stored trace and log excerpts to read for secrets. The signed report lists for each agent what its principal sees and which outside parties see it, and every secret found by kind and position without its value, and names exposure and blindness: principals who cannot see their own agent’s records, third parties and other principals who can with no stated basis, and secrets or credentials written into traces and logs (private keys, API keys, cloud and chat tokens, bearer tokens, signed web tokens and labeled passwords); then a principal who sees traces or logs only, outside parties with a stated basis and grants naming agents not in the list, with the fix for each. Whether stored memories are shared beyond their principal is wever_memory-sharing-check and whether a read-only scope holds is wever_read-only-scope-check. Verdict exposed_or_blind, access_gaps, access_unverified or access_sound. POST /api/observability-access-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_observability-alert-checkAgent Observability: submit the alert rules for an agent’s operation: the failure conditions that matter, each with its id, name and severity (critical, high, medium or low), and e…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Observability: submit the alert rules for an agent’s operation: the failure conditions that matter, each with its id, name and severity (critical, high, medium or low), and each alert with its id, the failure conditions it fires on, what it watches, its threshold, the period the threshold is read over, its owner, the owner’s next action or runbook, its channel and when it last fired in a test; and optionally the moment to read the rules at. The signed report lists every failure condition with the alerts that fire on it and every alert with its threshold read as numbered, vague or missing, and names alerting that would not work when it matters: failure conditions with no alert, alerts with no stated threshold, and alerts that name no owner and no next action; then an owner or a next action alone, thresholds with no number, no window, no test within 90 days, references to conditions not in the list and alerts tied to no condition, with the fix for each. Whether each production agent has a named owner, responder and fallback is wever_incident-ownership-check, when an agent hands a decision to a person is wever_escalation-threshold-check and routing delivered watch fires and job results is wever_alert-route. Verdict alerting_fails, alerting_gaps, alerting_unverified or alerting_works. POST /api/observability-alert-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_observability-coverage-checkAgent Observability: submit the list of an agent’s action types, each with the tools it uses and whether its runs are traced and logged; the tools the agent reaches, each with whet…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Observability: submit the list of an agent’s action types, each with the tools it uses and whether its runs are traced and logged; the tools the agent reaches, each with whether calls go through the traced path and the other ways the agent reaches it (a direct API key, a shell, a browser session); and either recent runs with whether a trace and log records were kept for each, or which runs keep traces and logs (all runs, a sample with its share, failures only, successes only, on request or none). The signed report lists each action, tool and run selection with its coverage, and names the blind spots: action types with no trace, tools reachable outside the traced path, and records kept for some runs but not others; then actions traced and not logged, tools an action uses that the list leaves out and reachable tools no action names, with the fix for each. Whether a gateway sits in front of every paid or external call is wever_gateway-coverage-check, which tools are exposed to an agent at all is wever_tool-exposure-check and whether one run’s trace reconstructs is wever_observability-trace-check. Verdict blind_spots, coverage_gaps, coverage_unverified or fully_covered. POST /api/observability-coverage-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_observability-retention-checkAgent Observability: submit a trace and log retention policy: each rule with the records it covers (traces, logs, an audit log, metrics, alerts, payment records or other), the runs…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Observability: submit a trace and log retention policy: each rule with the records it covers (traces, logs, an audit log, metrics, alerts, payment records or other), the runs it applies to (all runs, successful runs or failed runs), how long they are kept (a period such as 90 days or 7 years, or forever), how they leave when the period ends (deleted automatically, deleted by hand or not deleted) and whether they move to an archive first with how long the archive keeps them; and the dispute, audit or contract windows the work is under, each with its period and the records it needs. The signed report gives each rule how long its records stay available, each window the rules it reads, and failed runs beside successful ones, and names records that disappear too soon: retention shorter than the dispute or audit window the work is under, deletion with no archive, and failed runs losing their records sooner than successful ones; then failed runs kept longer, a rule for successful runs with none for failed runs, a window needing records no rule covers and traces or logs not kept at all, with the fix for each. A counterparty’s data retention and deletion policy is wever_data-retention-check and an agent memory setup’s retention is wever_memory-retention-check. Verdict records_lost, retention_gaps, retention_unverified or retention_holds. POST /api/observability-retention-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_observability-trace-checkAgent Observability: submit one run trace to find out whether the run can be reconstructed from it: each step with its id, its sequence number, its name, its kind (a tool call, a m…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,144 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t-observability-contract. Human page: https://weverlabs.com/agent-observability/."
Description, as the server wrote it Agent Observability: submit one run trace to find out whether the run can be reconstructed from it: each step with its id, its sequence number, its name, its kind (a tool call, a model call, a decision, a handoff, a message, a write or other), when it started and ended, the steps it follows from, the tool it called, whether its input and its result are recorded and its status; optionally the run with its task and start and end times; and the final outcome with its status, its text and the step it comes from. The signed report lists every step with its place in the run and whether its record is whole, and names what breaks reconstruction: no steps, steps another step reads from and numbers in the sequence that have no record, tool calls with no recorded result, steps with no timestamp or sequence number and steps sharing one timestamp, which leave the order of events unknowable, and a final outcome no recorded step leads to; then steps recorded before the steps they read from, calls with no recorded input, a succeeded outcome from a failed step, steps dated outside the run and steps that end before they start, with the fix for each. Classifying each recorded failure as recovered, worked around or fatal against the outcome is wever_trace-outcome-brief, a recovery attempt’s record is wever_recovery-record-check and incident claims against their timeline are wever_incident-evidence-check. A worked sample is wever_observability-trace-sample and the packet checklist is wever_observability-snapshot, both on https://weverlabs.com/mcp-free. Verdict trace_broken, trace_gaps, trace_unverified or trace_complete. POST /api/observability-trace-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-access-checkAgent Onboarding: submit every system and tool a new agent’s job needs or the agent can reach: optionally the agent with its id, when it runs its first task and where it runs its j…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
- Info
The description is 4,277 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
Description, as the server wrote it Agent Onboarding: submit every system and tool a new agent’s job needs or the agent can reach: optionally the agent with its id, when it runs its first task and where it runs its job (production, staging or a sandbox); each system with its id, name, kind (an API, a database, a tool, an inbox, a dashboard, a wallet, a file store or other), whether the written job needs it, whether the agent’s access is set up, for a system the job does not need whether access was removed or never granted, how access was tested (a real task run end to end, a connection only, or not tested), what the test showed (passed, failed or partly worked), where and when it was tested; and optionally the days before the first task after which a test counts as stale (14 when left out). The signed report gives every system with its test’s age in days and its status as proven, not_provisioned, failed, partial, elsewhere, stale, late, connection_only, untested, closed, open or unknown, and counts the systems, those needed, proven, broken and untested and those not needed and still open, and names broken access: a system the job needs with no access set up, and a failed test; then a test that only partly worked, one run only in another environment, one run too long before the first task or after it, and a system the job does not need left open, with the fix for each. A test of the connection only, no end-to-end test, and an unstated need, set-up, start, environment or test date leave the access untested. This reads the systems one new agent needs, provisioned and proven before its first task: which endpoints agents can call that delete, administer or move money is wever_tool-exposure-check, what a read-only agent can change is wever_read-only-scope-check, paths from a test environment into production systems are wever_sandbox-isolation-check and the rules a gateway enforces are wever_gateway-rules-check. Verdict onboarding_access_broken, onboarding_access_partial, onboarding_access_untested or onboarding_access_proven. POST /api/onboarding-access-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-credential-checkAgent Onboarding: submit the credentials a new agent was issued and the text its setup left behind: optionally the agent with its id and when it runs its first task; each credentia…8 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The title mentions "credential".
What this check looks forMatched text
title: "Onboarding credential check"
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "Agent Onboarding: submit the credentials a new agent was issued and the text its..."
- Info
The input schema at required[0] mentions "credentials".
What this check looks forMatched text
inputSchema.required[0]: "credentials"
- Info
The input schema at properties.credentials.items.properties.id.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.id.description: "The credential id, such as c1."
- Info
The input schema at properties.credentials.items.properties.name.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.name.description: "The credential, such as the orders API key."
- Info
The input schema at properties.credentials.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.description: "Each credential the agent was issued; empty when none."
- Info
The input schema at properties.texts.items.properties.text.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.texts.items.properties.text.description: "The text as written, read for credential-shaped strings; the report names each b..."
- Info
The description is 4,718 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
Description, as the server wrote it Agent Onboarding: submit the credentials a new agent was issued and the text its setup left behind: optionally the agent with its id and when it runs its first task; each credential with its id, name (such as the orders API key), the system it opens, its kind (an API key, a token, an OAuth grant, a password, a certificate, a signing key or other), what it can do there (read, write, admin or owner), what the written job needs there (none when the job does not need the system), where it is kept (a secrets manager, a vault, an environment variable, a config file, code, a prompt, a document, a chat, a log or not known), who it was issued to (this agent alone, several agents or people, a person or not known), when it expires and how often it is replaced (an empty list when none); and each prompt, log, document, ticket, chat, config or code text as written, read for credential-shaped strings such as a private key block, a provider secret key, a bearer token or a key assignment (an empty list when none). The signed report gives every credential with its status as secured, exposed, shared, unneeded, over_scoped, loose, expired, never_expires or unknown, every text with the kinds of credential found in it and its status as clean or credential_found, never repeating the string, and counts the credentials secured, exposed, shared, over-scoped and unneeded and the texts holding a credential, and names exposed credentials: one kept in code, a prompt, a document, a chat or a log, one found in supplied text, and one issued to several agents or to a person; then a level above what the job needs, a credential the job does not need, one kept in an environment variable or config file, one that expires before the first task and one with no expiry or rotation, with the fix for each. An unstated level, need, store or holder, or no supplied text, leaves the credentials unverified. This reads the credentials the agent itself was issued, not what a listing declares: the credentials, accounts and approvals a listing declares before payment are wever_credential-requirements-check, the credentials a test environment holds are wever_sandbox-isolation-check, what the credential behind each tool of a read-only agent can do is wever_read-only-scope-check and an operator’s key custody is wever_continuity-check. Verdict onboarding_credential_exposed, onboarding_credential_partial, onboarding_credential_unverified or onboarding_credential_secured. POST /api/onboarding-credential-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-identity-checkAgent Onboarding: submit a new agent’s identity at joining time: the agent with its id, the name people and logs see, the principal or owner it acts for, whether the owner confirme…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
- Info
The description is 4,650 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
Description, as the server wrote it Agent Onboarding: submit a new agent’s identity at joining time: the agent with its id, the name people and logs see, the principal or owner it acts for, whether the owner confirmed it acts for them and when it runs its first task; its identity with the identifier (such as a DID, a key id, a service account or a registry entry), its kind, who it was issued to (this agent, a person, a team, another agent or not known), how it was verified (a signature, a registry lookup, the issuer, the owner’s confirmation, the agent’s own word or not at all), when, and the identity record as written, read for wording that shares, borrows or hides the identity such as shared login, same key as or anonymous; and every agent already in the system with its id, name, identifier and owner (an empty list when none). The signed report gives the agent, the identity with any sharing wording and its status as established, not_its_own, unverified or unread, every other agent as same_identifier, same_name or distinct, and counts the other agents, the shared identifiers and shared names, and names a missing identity: no identifier of its own, one issued to a person, a team or another agent, one another agent already holds, wording that shares or hides it, and no owner or a placeholder owner such as the team; then an identifier never verified or resting on the agent’s own word, verification after the first task, an owner who has not confirmed the agent, a name another agent carries and no name, with the fix for each. An unstated issuer or verification, or no list of the other agents, leaves the identity unverifiable. This reads one new agent’s identity at joining time: one identity record’s own consistency before dealing with an outside agent is wever_agent-identity-check, whether an identity can be verified outside its issuing platform is wever_identity-portability-check, whether a registry covers every deployed agent is wever_agent-registry-check, whether the organisation keeps one current roster is wever_fleet-roster-check, the participants in one multi-agent job are wever_coordination-roster-check, a buyer acting for a principal is wever_delegated-identity-check and whether a handoff route checks its receiver is wever_receiver-identity-check. Verdict onboarding_identity_missing, onboarding_identity_partial, onboarding_identity_unverifiable or onboarding_identity_established. POST /api/onboarding-identity-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-permissions-checkAgent Onboarding: submit the permissions granted to a new agent and its written job: the job with its id, the written job as given to the agent, read for wording that hands over bl…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
- Info
The description is 4,430 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
Description, as the server wrote it Agent Onboarding: submit the permissions granted to a new agent and its written job: the job with its id, the written job as given to the agent, read for wording that hands over blanket access such as full access, whatever it needs or same access as, and each system it needs with its id, the system, the level it needs there (read, write or admin) and whether it needs it at all times or only during a task; and each grant with its id, the system it opens, the level (read, write, admin or owner), whether it stays on between tasks and how it was granted (directly, through a role or group, or copied from another agent’s or a person’s access). The signed report gives every grant with the level the job needs on its system and its status as matched, unneeded, above_need, below_need, standing, copied, unrated or unread, every need as granted, missing, under_granted or unrated, and counts the grants matched, unneeded, above the need and copied and the needs missing, and names excessive permissions: a grant on a system the written job does not list, admin or owner rights above the job’s need, a grant copied from another agent’s or a person’s access and blanket wording in the written job; then write where the job only reads, a standing grant the job needs only during a task, a need with no grant and a grant below the need, with the fix for each. No list of needs or an unstated level leaves the permissions unchecked. This reads the permissions granted at joining against the written job, not one action’s authority: whether an actor’s access fits its authority grant for an action is wever_authority-scope-check, a delegation’s written bounds are wever_delegation-scope-check, a stored grant evaluated for an exact subject, action and resource is wever_agent-permission-ledger, a trading or treasury mandate is issued and read with the mandate tools such as wever_mandate-status, which endpoints agents can call is wever_tool-exposure-check and which permissions rest on the model is wever_enforcement-layer-check. Verdict onboarding_permissions_excessive, onboarding_permissions_partial, onboarding_permissions_unchecked or onboarding_permissions_matched. POST /api/onboarding-permissions-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-record-checkAgent Onboarding: submit the record a new agent’s onboarding leaves behind and what is live: optionally the agent with its id and name; the record with its id, who approved the set…5 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...ers API read) and kind (a permission, a credential, a wallet, a limit, a system or other),..."
- Info
The input schema at properties.record.properties.granted.items.properties.kind.enum[1] mentions "credential".
What this check looks forMatched text
inputSchema.properties.record.properties.granted.items.properties.kind.enum[1]: "credential"
- Info
The input schema at properties.record.properties.granted.items.properties.kind.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.record.properties.granted.items.properties.kind.description: "A permission, a credential, a wallet, a limit, a system or other."
- Info
The input schema at properties.live.items.properties.kind.enum[1] mentions "credential".
What this check looks forMatched text
inputSchema.properties.live.items.properties.kind.enum[1]: "credential"
- Info
The description is 4,178 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
Description, as the server wrote it Agent Onboarding: submit the record a new agent’s onboarding leaves behind and what is live: optionally the agent with its id and name; the record with its id, who approved the setup, when, when the agent’s first task ran, everything it says was granted, each with its item (such as orders API read) and kind (a permission, a credential, a wallet, a limit, a system or other), when the setup is next reviewed and who reviews it; everything actually live for the agent now, read from its systems, each with its item and kind (an empty list when nothing is); and when the live state was read. The signed report gives the record, every recorded item as live, not_live or unread, every live item as recorded, unrecorded or unread, and counts the items recorded, live, recorded and live, live but unrecorded and recorded but no longer live, and names a missing record: no approver, approval date or grant list, approval by the agent itself or by a placeholder, approval after the first task ran and a live item missing from the record; then a recorded item no longer live, no review date, an overdue review and no named reviewer, with the fix for each. No live state, no read date and no first-task date leave the record unverifiable. This reads the record one onboarding leaves behind against what is live: whether a registry covers every deployed agent is wever_agent-registry-check, whether one roster is kept current as agents come and go is wever_fleet-roster-check, whether an approval record holds the exact call is wever_approval-record-check, whether recorded actions can be attributed to an agent under an authorization is wever_actor-attribution-check, stored permission grants and revocations are wever_agent-permission-ledger and a consent’s ledger entry is wever_consent-record-check. Verdict onboarding_record_missing, onboarding_record_incomplete, onboarding_record_unverifiable or onboarding_record_complete. POST /api/onboarding-record-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_onboarding-wallet-checkAgent Onboarding: submit the wallet a new agent is set up with, before it needs one: optionally the agent with its id and when it runs its first task; the job with whether it moves…2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
- Info
The description is 4,549 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
Description, as the server wrote it Agent Onboarding: submit the wallet a new agent is set up with, before it needs one: optionally the agent with its id and when it runs its first task; the job with whether it moves money, the most it needs the wallet to hold (such as the first month’s spend), its unit, the network it pays on and the asset it pays in; the wallet with its id or address, its kind (a wallet for this agent alone, one shared with other agents, the owner’s own wallet, the treasury or not known), network, asset, balance, unit, when it was set up and when funds first moved into it; the recovery with who recorded it (the owner, an operator, the agent itself, nobody or not known), when, how the wallet is recovered and the recovery note as written, read for a seed phrase or key written out in it; and optionally the share above the job’s need at which a balance counts as overfunded (20 when left out). The signed report gives the wallet with its balance as a share of what the job needs, the recovery with whether it was written out and whether it came before funding, every requirement (a wallet for this agent alone, funded to what the job needs, on the job’s network and asset, recovery recorded by the owner, recovery recorded before funds move, set up before the first task) as met, unmet or unknown, and counts them, and names an unsafe wallet: the agent drawing on a shared, owner or treasury wallet, no recovery, a recovery held by the agent, funds moved in before the recovery was recorded and a seed phrase or key written out in the note; then a balance past the headroom, another network or asset, a wallet set up or funded after the first task, an empty wallet and a recovery recorded by an operator, with the fix for each. When the job moves no money, no funds should be held. This reads one new agent’s wallet before funds move, beside the wallet tools already in the catalog: a month of an address’s activity is wever_wallet-statement and its recent flows wever_wallet-activity-brief, concentration caps across balances are wever_treasury-policy-check, an operator’s key recovery is wever_recovery-readiness-check and a budget’s five controls are wever_budget-controls-check. Verdict onboarding_wallet_unsafe, onboarding_wallet_partial, onboarding_wallet_unverified or onboarding_wallet_ready. POST /api/onboarding-wallet-check. Reads only the recorded onboarding material in the call: no network request, fetch, money movement, message, configuration change or external write; no credential is issued, rotated or revoked, no permission granted or removed, no wallet funded or moved, no limit set and the agent is not approved, and a credential found in supplied text is named by kind and never repeated. Agent Onboarding reads the setup of a new agent joining a system, before its first task: whether its identity is established and tied to its owner, whether its credentials are scoped to its job, stored where they belong and kept out of prompts, logs and shared documents, whether its wallet is the right one, funded only to the job with recovery recorded by the owner before funds move, whether its permissions match the written job, whether its limits sit in code with a human handoff, whether the systems it needs are provisioned and tested, whether it is ready for its first task and whether the onboarding record matches what is live. It is not vetting an outside party (wever_counterparty-onboarding-check reads a counterparty’s onboarding packet before the system transacts with it; an onboarding check reads an agent joining the system itself), not a grant or an authority for one action (wever_consent-scope-check reads whether a principal’s grant covers a use and wever_authority-scope-check whether an actor’s access fits its grant for an action; an onboarding check reads that the grant and the authority were put in place before the first task, and does not re-check them per action), not a limit in flight (the wever_limit checks and wever_agent-budget-guard read limits for work already running; an onboarding check reads that limits were set in code before it starts) and not an action’s consequences (the wever_risk checks read one planned action; an onboarding check reads readiness to start). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_ownership-map-briefAgent Oversight: submit the work items being handed to agents (outcome, agent, owner and owner type where known), optional known people with load limits, known agent names and a lo…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit the work items being handed to agents (outcome, agent, owner and owner type where known), optional known people with load limits, known agent names and a load limit per owner. The signed brief maps the accountable person for each outcome and flags unowned items, items owned by an agent or a group rather than a person, and owners carrying more than the stated load. A supplied handover package is wever_handover-check. Verdict unowned_outcomes, agent_owned_outcomes, owners_overloaded or all_owned. POST /api/ownership-map-brief. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_paid-access-checkAgent Operations: submit one purchase record for paid content or access (what was bought, price, date, payment reference, the paying agent and the person it bought for) with the se…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Operations: submit one purchase record for paid content or access (what was bought, price, date, payment reference, the paying agent and the person it bought for) with the seller’s access terms as supplied and, where useful, a date to hold the access against. The signed report states what the buyer holds: whether access continues after purchase and for how long, whether it extends to the buyer’s human or only the paying agent where the terms say, the conditions on it, and the proof to keep to re-establish access later. Terms analysis on supplied records; the content is never retrieved and the seller never contacted. Verdict access_ended, terms_unclear, access_limited or access_continues. POST /api/paid-access-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_partial-state-checkAgent Recovery: submit a task system’s state reporting: whether it has a state for partly done work apart from done and failed, whether a partly done task lists the pieces that did…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Recovery: submit a task system’s state reporting: whether it has a state for partly done work apart from done and failed, whether a partly done task lists the pieces that did not complete, what lets a task be reported done (every part confirmed, the last step returning, no error raised or any response) and what it reports when a task fails partway (partly done, failed, done, still in progress or nothing), plus optionally the tasks it reported on after a mid-task failure with each piece, whether it finished and the pieces the report named as missing. The signed report says whether partial completion is visible and labeled as partial with its missing pieces named, and names a system that reports done on a mid-task failure, on any response or on no error, or reports nothing, and every task reported done with unfinished pieces (half-done work presenting as finished), then no partial state, done on the last step alone, a partial failure shown as failed or still running, missing pieces left unnamed and every task reported failed with finished pieces, with the fix for each. This reads what an interrupted task’s reported state shows across a system: one paid call’s delivered result against its order is wever_result-validation-check, what a stop does to in-flight actions is wever_kill-switch-check and where an interrupted task picks up is wever_resume-point-check. Verdict half_done_looks_done, partial_state_gaps, partial_state_unverified or partial_shown_as_partial. POST /api/partial-state-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_policy-briefDiligenceOps: submit one policy excerpt, such as an insurance, warranty or service policy.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit one policy excerpt, such as an insurance, warranty or service policy. The signed brief states the coverage, the exclusions as items, the limits by kind (per claim, aggregate, per item, deductible, copay) with their amounts, and the conditions a claim must meet with their time limits, each quoted, and lists what the excerpt does not state. POST /api/policy-brief. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_policy-precedence-checkAgent Oversight: submit a standing policy set in advance (per-action and period limits, allowed and blocked counterparties, who may approve exceptions), a momentary approval (appro…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit a standing policy set in advance (per-action and period limits, allowed and blocked counterparties, who may approve exceptions), a momentary approval (approver, amount, counterparty, expiry) and the action taken or proposed. The signed report says which instruction governs and why, whether the action sat inside both, and the exact point of conflict when they disagree. Standing limits are part of the instruction. An action against policy text is wever_policy-fit-check. Verdict outside_standing_policy, outside_approval or inside_both. POST /api/policy-precedence-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_policy-replay-checkJob Records: submit a proposed policy change as structured rules (per-action caps, period caps per day, week, month or in total, allow lists, block lists and approval thresholds, e…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Job Records: submit a proposed policy change as structured rules (per-action caps, period caps per day, week, month or in total, allow lists, block lists and approval thresholds, each optionally limited to some counterparties, categories or services) with a batch of past job traces or payment records. The signed report replays the rules over the history in time order and states, per rule, which past actions would have been blocked, which would have needed approval and which pass unchanged, with counts and every affected record named. Supplied history only: a live x402 challenge is wever_payment-policy-check. Verdict would_block, would_need_approval or no_change. POST /api/policy-replay-check. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_post-jobPost a job with a fixed USDC fee, deliverable and acceptance criteria, bound to the hiring agent and its human principal.Reads text others writeCan send data outNo findings
Why it is in this group
- The name "wever_post-job" says "post": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe name "wever_post-job" says "post": it sends or publishes something to other people.
Description, as the server wrote it Post a job with a fixed USDC fee, deliverable and acceptance criteria, bound to the hiring agent and its human principal. POST /api/post-job. 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. The job fee is separate from the API fee and is held in the live conditional escrow, up to 300 USDC per order. Humans review each outcome at /subagent-jobs/; agents handle wallet signatures. Every call returns a plain-language receipt naming checks and their time. Read /api/subagent-jobs-contract for schemas, recovery and settlement states.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_price-truth-checkAgent Purchases: submit the quote an agent received for one paid call (amount, asset, network, quantity, unit price, what counts as delivered, when it was quoted and when it lapses…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Purchases: submit the quote an agent received for one paid call (amount, asset, network, quantity, unit price, what counts as delivered, when it was quoted and when it lapses) and the charge settled for it (amount, asset, network, quantity charged, quantity delivered, when it settled and its reference). The signed report compares the amount, asset, network and quantity field by field and names every mismatch: an amount that moved between quote and settlement in either direction with the difference and its share of the quote, a different asset or network, a different quantity and units charged but not delivered, then a quote whose unit price and quantity do not add up and a charge settled after the quote lapsed. Quoted terms against governing published terms are wever_terms-match-check, a live 402 challenge against advertised terms is wever_endpoint-honesty-check, charges against a fee schedule are wever_fee-reconciliation-check and an invoice against a payment is wever_invoice-payment-match. Verdict price_mismatch, quote_gaps, price_unverified or price_true. POST /api/price-truth-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_pricing-evidence-checkDiligenceOps: submit a pricing page (text, or a public HTTPS url).Reads text others writeCan send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it DiligenceOps: submit a pricing page (text, or a public HTTPS url). The signed report lists the tiers with their prices (amount, currency, period, seat basis and billing cycle), limits and features, plus other priced items, and flags the same tier or item priced two ways on the same basis and a limit a tier states with two different values, each quoted with its offset. POST /api/pricing-evidence-check. Reads the supplied text, or one public HTTPS page read once with GET (no redirects, public DNS only, 512 KB): no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_pricing-tier-checkDiligenceOps: submit a pricing page’s tier descriptions, as text or as tiers with price, limits and features, and the plan terms.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a pricing page’s tier descriptions, as text or as tiers with price, limits and features, and the plan terms. The signed report reads each tier’s price, limits and features from sections or tier columns and lists the prices, limits (another figure, or a cap on an unlimited one) and features (excluded, a paid add-on, or named for another tier) that differ between the page and the terms, with quotes. POST /api/pricing-tier-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_privacy-evidence-checkDiligenceOps: submit privacy claims and the privacy policy or other supporting document.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit privacy claims and the privacy policy or other supporting document. The signed report tags each claim collection, sharing or retention and marks it supported, contradicted or never mentioned by the document, quoting the sentences behind each finding, such as a retention period or a sharing practice that differs. POST /api/privacy-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_proof-relayPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...POST /api/proof-relay. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Prepare a reference envelope and a digest binding the supplied source and destination, ready for the sender’s delivery step. Operating boundary: Supply source, destination and artifact references. Read the prepared envelope and binding digest. The sending agent delivers it through its chosen transport; the human reviews the references and resulting acknowledgment. POST /api/proof-relay. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_provider-readiness-checkAgent Purchases: submit a provider’s public purchase facts as an agent buyer meets them: each fact it discloses (the price, the asset and network, how a quote holds, what counts as…Can send data out1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a URL (input.provider.endpoint): a request to a URL can carry data to whoever runs it.
Findings
- Info
The description is 2,075 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/."
Description, as the server wrote it Agent Purchases: submit a provider’s public purchase facts as an agent buyer meets them: each fact it discloses (the price, the asset and network, how a quote holds, what counts as delivered, how a call can be rehearsed, the refund route and the terms) with when the buyer meets it (before payment, in the 402 challenge, after payment or not at all) and whether it is stated or only implied, how a call is priced (fixed, per unit, quoted or variable, with whether a quote binds the amount settled, how long it holds and any cap), the rehearsal it offers (dry run, sandbox, simulated response, testnet or none, whether it is free and whether it runs on the paid call’s own interface) and how money moves back (refund endpoint, refund process, automatic, contact-us or none, with its window). The signed report gives each readiness piece (disclosure before payment, price and quote, rehearsal, money back) as ready, implied, missing or unstated, and names every fact shown only after payment or not at all, every quote that does not bind, every uncapped variable price, a missing or paid rehearsal and a missing or contact-only refund route, with the fix for each. Readiness for recovery, payouts and adoption is wever_recovery-readiness-check, wever_payout-readiness-check and wever_adoption-readiness-check, and an anonymous visitor’s view of a whole public layer is wever_public-surface-check. A summary sample is wever_provider-readiness-sample on https://weverlabs.com/mcp-free. Verdict not_ready, implied_readiness, readiness_unverified or ready_for_agent_buyers. POST /api/provider-readiness-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_public-surface-checkAgent Access: submit a site’s public layer as an anonymous visiting agent meets it: every entry point (manifest, llms.txt, agent card, OpenAPI description, MCP server, tool, page, …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Access: submit a site’s public layer as an anonymous visiting agent meets it: every entry point (manifest, llms.txt, agent card, OpenAPI description, MCP server, tool, page, endpoint or feed) with the authentication it needs (none, optional or required), whether it is machine readable, what it tells a visitor (offering, next step, authentication requirements, terms, contact, pricing) and what a visitor can do or see through it (describe, read, search, quote, transact, write, personal or account data). The signed report names every discovery gap, where an anonymous agent cannot find a machine-readable entry, tell what is offered, learn how to proceed or learn what authentication the transactional layer needs, and every leak, where a transaction, a write or personal or account data is reachable without authentication. One agent card’s completeness is wever_agent-card-check and an offer across every discovery surface is wever_discovery-coverage-check. Verdict leaks_found, discovery_gaps, surface_unverified or discoverable_and_contained. POST /api/public-surface-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_receipt-verifyDiligenceOps: verify a Wever signed receipt against Wever’s published Ed25519 keys and state what it proves, or exactly why it fails.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: verify a Wever signed receipt against Wever’s published Ed25519 keys and state what it proves, or exactly why it fails. Accepts the compact token or the receipt object, plus the optional report it came with. Covers data evidence, Wever Watch, DiligenceOps, receipt passport, handoff pack and scheduled observation receipts. Runs ordered checks (format, algorithm, family, key id, encoding, signature, type, issuer, receipt fields, report digest) and names the first that fails. POST /api/receipt-verify. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-contract. Human page: https://weverlabs.com/diligence/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_receiver-identity-checkAgent Handoffs: submit a handoff addressing setup: the registered agents with each registry record and registered endpoint, and each handoff route with the agent it means to reach,…3 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...registry record, by its signature, its credential or mutual TLS), when (before the task m..."
- Info
The input schema at properties.routes.items.properties.verified_by.items.enum[2] mentions "credential".
What this check looks forMatched text
inputSchema.properties.routes.items.properties.verified_by.items.enum[2]: "credential"
- Info
The input schema at properties.routes.items.properties.verified_by.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.routes.items.properties.verified_by.description: "...gistry record, by its signature, by its credential or by mutual TLS."
Description, as the server wrote it Agent Handoffs: submit a handoff addressing setup: the registered agents with each registry record and registered endpoint, and each handoff route with the agent it means to reach, how the sender addresses it (its registered name resolved through the registry, a fixed endpoint, discovery, the reply-to address in a message, or whichever endpoint answers), the address used, how the receiver’s identity is checked (against its registry record, by its signature, its credential or mutual TLS), when (before the task moves, after or never) and what the sender does on a mismatch. The signed report names every route where any endpoint that answers can receive the task (sent to whichever endpoint answers, unchecked, checked only after the task moves, or going ahead on a mismatch), then every receiver missing from the registry, every address its record does not list, every check that never consults the registry record and every mismatch that only warns, with the fix for each. This reads whether each handoff route checks its receiver against its registered record before work moves: one identity record’s own consistency is wever_agent-identity-check, whether a registry covers the fleet is wever_agent-registry-check and a buyer acting for a human principal is wever_delegated-identity-check. Verdict any_endpoint_receives, identity_gaps, identity_unverified or receiver_verified. POST /api/receiver-identity-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-handoffs-contract. Human page: https://weverlabs.com/agent-handoffs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_recommendation-disclosure-checkJob Records: submit one recommendation (a product or service suggestion with its text) with its commercial terms as they stand (affiliate commission with its rate or amount, paid p…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Job Records: submit one recommendation (a product or service suggestion with its text) with its commercial terms as they stand (affiliate commission with its rate or amount, paid placement, sponsorship, or none) and where it will appear. The signed report states whether the commercial interest is disclosed in the recommendation itself, whether the disclosure names the right kind and matches the supplied rate or amount, whether it comes before the recommendation, and the exact disclosure line the terms call for when it is missing or understated. Disclosure only; the merit of the recommendation is outside the report. Verdict not_disclosed, understated, disclosed or nothing_to_disclose. POST /api/recommendation-disclosure-check. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_recovery-readiness-checkAgent preflight round 2: check recovery arrangements before you depend on them.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight round 2: check recovery arrangements before you depend on them. Input text, fields (backups, key_recovery, export_paths, restore_steps, restore_test) or both. The signed report marks backups and their recency, the key recovery method, export paths for records and catalogs, restore steps and evidence the restore was tested ready, partial or missing, with quotes, and names the weakest links plainly, missing first. POST /api/recovery-readiness-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_recovery-record-checkAgent Recovery: submit a system’s recovery records (each with its id, the task interrupted, when, what failed, what was done: resumed, restarted, rolled back or abandoned, where a …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Recovery: submit a system’s recovery records (each with its id, the task interrupted, when, what failed, what was done: resumed, restarted, rolled back or abandoned, where a resume picked up, what was lost, what changed afterward and when the record was written) and optionally the interruptions seen in other logs, such as alerts, error logs or a tracker. The signed report counts every recovery that keeps the essentials and names every record missing what failed, what was done or what was lost (the core a recovery cannot be reconstructed without), every record missing what changed afterward, every record written more than a day after the interruption, and every interruption seen elsewhere with no recovery record behind it, with the fix for each. This reads the record of interruptions and recoveries specifically: the record of limit events is wever_limit-evidence-check, the buyer’s record of paid calls is wever_purchase-record-check and incident claims against a timeline are wever_incident-evidence-check. Verdict recoveries_unrecorded, record_gaps, records_unverified or recoveries_recorded. POST /api/recovery-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_rehearsal-checkAgent Purchases: submit one paid call (name, endpoint, amount, network and the fields its result carries) and every rehearsal the provider offers for it: dry runs, sandboxes, simul…Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a URL (input.call.endpoint): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Agent Purchases: submit one paid call (name, endpoint, amount, network and the fields its result carries) and every rehearsal the provider offers for it: dry runs, sandboxes, simulated responses, testnet and free calls, each with its endpoint, its cost, the price it shows (the live quote, a fixed sample or none), how much of the payment step it runs (the full 402 challenge, a signature only or none), what comes back (live data, the same schema, an example or none) and its fields, whether it runs at production speed, whether it can be made to fail and its network. The signed report says, for the price, the payment handshake, the result shape, timing, failure paths and the network, which rehearsal proves it and why each other falls short, and names every dimension the first paid call would be the first to test, plus rehearsals that cost money or run against another endpoint. Whether a provider offers a rehearsal at all is wever_provider-readiness-check and whether a test environment reaches production is wever_sandbox-isolation-check. Verdict no_rehearsal, rehearsal_gaps, rehearsal_unverified or rehearsal_covers_call. POST /api/rehearsal-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_renewal-notice-checkCommerce Assurance: submit agreement text, renewal fields or both, with the current date.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Commerce Assurance: submit agreement text, renewal fields or both, with the current date. The signed report states the renewal type, the current term rolled forward through automatic renewals, the notice period and the notice deadline computed from the dates, price-change terms and the termination window, with ambiguous or missing renewal language flagged. Verdict notice_window_open, notice_deadline_passed, no_automatic_renewal or renewal_terms_unclear. Other document deadlines are wever_deadline-check. POST /api/renewal-notice-check. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-decay-checkAgent Reputation: submit one reputation score’s aging rules: the headline score with the window it states (such as 12 months or all time), the aging method (none, a window, linear …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Reputation: submit one reputation score’s aging rules: the headline score with the window it states (such as 12 months or all time), the aging method (none, a window, linear across a window, a half-life or steps) with its window, half-life or the weight from each age on, and optionally the entries the score reads with each one’s date and outcome, the recent period read against older volume (90 days when left out) and the moment to read it at. The signed report gives the weight a result keeps at one month, six months and one, two and three years, the recent failure rate beside the weighted failure rate the score reads and the share of the weight older jobs carry, and names a reputation that never ages: years-old results weighted the same as last month’s, no window stated for the headline score and recent failures diluted by old volume, then a decay so slow that a two-year-old result keeps half its weight, an all-time window, a headline window that differs from the rule’s, an incomplete rule and undated entries, with the fix for each. Two evaluation runs compared for drift are wever_evaluation-drift-check, prior answers against a later change are wever_answer-freshness-check and a trust score computed from verified work is wever_agent-trust-scorecard. Verdict never_ages, aging_gaps, aging_unverified or ages_with_time. POST /api/reputation-decay-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-dispute-checkAgent Reputation: submit how one reputation record handles disputes: each dispute the record shows with its id, job, when it was opened, its status (open, resolved or withdrawn), i…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Reputation: submit how one reputation record handles disputes: each dispute the record shows with its id, job, when it was opened, its status (open, resolved or withdrawn), its outcome (for the agent, for the counterparty, split, refunded or withdrawn) and when it was resolved; the disputes on file in the dispute log, escrow or platform records the record draws on, in the same shape; the dispute counts the record states (total, open and resolved); and how much each kind of dispute moves the score (open, resolved for the agent, resolved against it and withdrawn). The signed report matches every dispute shown to the log and counts the disputes shown by status, and names concealment: disputes on file missing from the record, disputes shown without their outcomes, statuses or outcomes shown differently from the log and resolved disputes counted the same as open ones, then stated counts that differ from the disputes shown, undated disputes and resolutions and disputes linked to no job, with the fix for each. One Wever dispute with its signed receipt chain is wever_dispute-status, incident claims against their timeline are wever_incident-evidence-check and the record’s own entries and totals are wever_reputation-record-check. Verdict disputes_concealed, dispute_gaps, disputes_unverified or disputes_disclosed. POST /api/reputation-dispute-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-gaming-checkAgent Reputation: submit one reputation system’s design: how identities are created, with what a new identity costs, what it proves (nothing, an email, a wallet, KYC, an attestatio…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Reputation: submit one reputation system’s design: how identities are created, with what a new identity costs, what it proves (nothing, an email, a wallet, KYC, an attestation or a stake), where its score starts (zero, neutral, inherited, purchased or imported) and whether a score moves to another identity or can be sold; how ratings are earned, with who can rate (anyone, any account, a counterparty or a counterparty with settled work), whether a rating needs a settled job, the smallest job that earns one and how ratings are weighted; and what the design detects: pairs that rate each other, closed rating rings, raters funded from one source and an agent rated by its own operator or wallets. The signed report lists each gaming path as open or closed with what a rating costs to fake, and names a fresh identity starting with a borrowed or bought score, ratings from parties with no settled work behind them and review exchanges the design cannot see, then free identities, ratings that cost only the smallest job, unseen rating rings, unseen self-dealing or shared funding and unweighted ratings, with the fix for each. Headline payment numbers with no commerce behind them are wever_volume-quality-check, quoted endorsements against their source are wever_testimonial-evidence-check and one agent identity record’s consistency is wever_agent-identity-check. Verdict gameable, gaming_gaps, gaming_unverified or gaming_resistant. POST /api/reputation-gaming-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-portability-checkAgent Reputation: submit how one reputation record is stored and presented: the issuing platform, the agent, who holds the record (the issuer’s account, the agent, both or a third …2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...ts, its format (JSON, CSV, a verifiable credential, signed attestations, on chain, PDF, HT..."
- Info
The description is 2,152 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/."
Description, as the server wrote it Agent Reputation: submit how one reputation record is stored and presented: the issuing platform, the agent, who holds the record (the issuer’s account, the agent, both or a third party) and what it is tied to (a platform account, the agent’s key, a DID, a wallet, a domain or other); its export, with whether one exists, its format (JSON, CSV, a verifiable credential, signed attestations, on chain, PDF, HTML, proprietary or none), whether the agent exports it on its own and what it carries (entries, dates, counterparties, outcomes, ratings, disputes and signatures); how another party verifies an entry (a signature, on chain, the issuer’s API, a public page or nothing), whether that works with no account at the issuer and whether the key is published; whether the record survives account closure and whether the issuer can rewrite it; and the platforms beyond the issuer that accept it. The signed report says whether the record moves with the agent and names a record locked to one platform account, no export format and nothing verifiable outside the issuing platform, then an export another platform does not read as data, an export missing parts of the record, an export only on request, an unpublished signing key, a record lost on account closure, entries the issuer can rewrite and no other platform accepting it, with the fix for each. An agent identity setup’s portability is wever_identity-portability-check, an agent memory export is wever_memory-export-check and a signed passport for completed Wever computation is wever_receipt-passport. Verdict platform_locked, portability_gaps, portability_unverified or portable. POST /api/reputation-portability-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-record-checkAgent Reputation: submit one agent’s reputation record: each entry with its id, the job record it rests on, when the job was completed, the counterparty, the outcome (completed, pa…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Reputation: submit one agent’s reputation record: each entry with its id, the job record it rests on, when the job was completed, the counterparty, the outcome (completed, partial, failed, cancelled or refunded), the rating, the amount and whether a dispute was raised; the job records themselves with their counterparty, date and amount; the totals the record states (jobs, completed, failed, disputed, average rating and total value); the rating scale; and optionally the moment to read it at. The signed report reads the record underneath a score entry by entry, says for each whether it rests on a job record and recomputes every total from the entries listed, and names no entries, entries with no date, entries with no counterparty, outcomes with no underlying job record and totals that do not match the entries listed, then entries with no outcome, entries that differ from their job record, job records left out of the record, entries dated after the read moment and ratings above the scale, with the fix for each. Computing a trust score from verified completed work is wever_agent-trust-scorecard, one job’s payments against its receipts is wever_job-receipt-pack and headline payment numbers with no commerce behind them are wever_volume-quality-check. A worked sample is wever_reputation-profile-sample and the packet checklist is wever_reputation-snapshot, both on https://weverlabs.com/mcp-free. Verdict record_unreliable, record_gaps, record_unverified or record_holds. POST /api/reputation-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_reputation-source-checkAgent Reputation: submit the sources behind one reputation record or score: each source with its id, kind (self-reported, a platform record, a counterparty attestation, a third-par…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,049 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/."
Description, as the server wrote it Agent Reputation: submit the sources behind one reputation record or score: each source with its id, kind (self-reported, a platform record, a counterparty attestation, a third-party audit, an on-chain record, a payment receipt or other), its issuer, whether it is independent of the agent and its operator, whether a third party can check it, when it was produced or last confirmed and how the record shows it (as verified, as self-reported or unlabeled); each claim with the source ids it rests on, whether it is presented as current and how it is shown; and optionally the subject agent and headline score, how old a source may be and still carry a current claim (12 months when left out) and the moment to read it at. The signed report says source by source whether it is self-reported, independent, checkable and how old it is, and claim by claim what it rests on, and names self-reported history presented as verified, a single source behind every claim, current claims resting only on sources too old to support them and claims with no source, then sources shown as verified that nobody else can check, claims resting only on parties tied to the agent, undated sources and self-reported history shown with no label, with the fix for each. Quoted endorsements against their source text are wever_testimonial-evidence-check, one claim against supplied evidence is wever_claim-evidence-check and a one-call brief on a counterparty is wever_counterparty-brief. Verdict sourcing_weak, sourcing_gaps, sourcing_unverified or sourcing_sound. POST /api/reputation-source-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_result-validation-checkAgent Purchases: submit the order’s own terms for one paid call (the values it asked for, the fields a delivered result carries, what counts as delivered, how fresh the data must b…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Purchases: submit the order’s own terms for one paid call (the values it asked for, the fields a delivered result carries, what counts as delivered, how fresh the data must be and the fewest items that count) and the result that came back (HTTP status, its fields with their values, its item count, when its data was current, when it arrived and any error carried in its body). The signed report checks each term met, not met or unstated and names every failure: a non-2xx status, an error inside a success, a requested value the result carries differently, a required field it lacks, an empty or short result and stale data. A 200 response carrying the wrong data is reported as a silent failure, not a success. A job’s work against acceptance criteria is wever_deliverable-check and a result that never arrived after payment is wever_payment-recovery-check. Verdict silent_failure, order_not_met, validation_unverified or order_met. POST /api/result-validation-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_resume-point-checkAgent Recovery: submit a task system: each step of the task in order with whether finishing it writes a record a resumed run reads and what running it again costs, what the system …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Recovery: submit a task system: each step of the task in order with whether finishing it writes a record a resumed run reads and what running it again costs, what the system does after an interruption (resumes from the recorded point, restarts from the beginning or leaves it to an operator), where the progress record lives (a durable store, the running process, the agent’s context or nowhere) and whether it names the steps done and the steps that remain and keeps what finished steps produced, plus optionally the interruptions seen with the step each stopped in and where the next run picked up. The signed report says whether an interrupted task resumes from a recorded point or starts over, and names a system that restarts, keeps no record, keeps one an interruption wipes or does not name what is done, and every run seen starting again from the beginning, then a record missing what remains or the outputs, every step that records nothing, an operator deciding each time and every run that redid finished steps, with the cost re-spent and the fix for each. This reads where interrupted work picks up across a system: one half-finished job’s state assembled into a resume pack is wever_job-handoff-pack, backups and restore steps for data are wever_recovery-readiness-check and a result that never arrived after payment is wever_payment-recovery-check. Verdict work_restarts, resume_gaps, resume_unverified or resumes_from_record. POST /api/resume-point-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_retry-plan-checkAgent Recovery: submit a system’s retry behavior: the standing retry plan (whether it is written down, the most attempts, how attempts are spaced, the first and longest wait, how l…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,076 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/."
Description, as the server wrote it Agent Recovery: submit a system’s retry behavior: the standing retry plan (whether it is written down, the most attempts, how attempts are spaced, the first and longest wait, how long retrying may go on, the failures never retried, what happens when retrying stops and who hears, and whether rate-limit waits are honored), optionally the calls that retry with whether each follows the plan, its own attempt limit and whether it charges, sends, books or writes, and the failures the system sees with whether retrying can fix each and whether it is retried. The signed report states the plan’s headline (attempts, spacing and stop point) and names no plan, no bound on attempts or time, a stop nobody hears of, every failure retrying cannot fix that is retried and every call retrying without limit outside the plan, then a plan set only in code, retries at once, growing waits with no jitter, no never-retry list, a stop only logged or told to no one named, rate-limit waits ignored, every call on its own terms and every state-changing call retried, with the fix for each. This reads the standing plan across calls: whether one retry of a specific payment or refund flow can charge twice is wever_retry-safety-check, whether a resumed or retried task repeats a message, order, record or booking is wever_duplicate-effect-check and where a handed-off task goes when its receiver fails is wever_handoff-failure-check. A summary sample is wever_retry-plan-sample on https://weverlabs.com/mcp-free. Verdict retries_blind, retry_plan_gaps, retry_plan_unverified or retry_plan_stated. POST /api/retry-plan-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_retry-safety-checkAgent Readiness: submit a payment or refund flow’s retry design as paths (operation, timeout, how long the provider can take, retry triggers, most retries, idempotency key and its …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Readiness: submit a payment or refund flow’s retry design as paths (operation, timeout, how long the provider can take, retry triggers, most retries, idempotency key and its scope, duplicate detection and its window, status check before retry). The signed report names every path where a retry can produce a second charge, refund or settlement, the paths protected by an idempotency key, a confirmation check or duplicate detection, and the smallest fix for each unprotected path. Design review of the supplied material; live order records are wever_duplicate-order-check. Verdict double_effect_possible, protection_unverified or retry_safe. POST /api/retry-safety-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-blast-radius-checkAgent Risk: submit one planned action and everything it can touch if it goes wrong: the action with its id, what it is meant to touch (such as one customer account) and its descrip…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,850 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
Description, as the server wrote it Agent Risk: submit one planned action and everything it can touch if it goes wrong: the action with its id, what it is meant to touch (such as one customer account) and its description, read for wording that selects everything such as all users, every account or bulk; each item it can touch with its id, name, kind (accounts, users, systems, funds or data), how it relates to the target (the target itself, something the target shares with others, something fed by the action, or not known), how many it covers, whether a boundary keeps a failure from reaching it (such as a per-account key, a separate wallet or a sandbox) and that boundary as written; how many items the action runs over in one go (1 when left out); and whether it stops at the first failure. The signed report gives the action with any select-everything wording, every item with its status as target, contained, exposed or unmapped, and counts the items at the target, beyond it, contained, exposed and unmapped with how many accounts, users or records the exposed items cover, and names a wide blast: wording that selects everything, and a failure reaching funds, accounts or users beyond the target with no boundary; then a failure reaching shared systems or data, a run over many items that carries on after a failure and a boundary named but not written down, with the fix for each. An item whose relation or boundary is not stated is unmapped. This reads how far one action’s failure spreads beyond its target: which endpoints agents can call that delete, administer or move money is wever_tool-exposure-check, paths from a test environment into production systems or data are wever_sandbox-isolation-check and wever_escape-path-check, and what a kill switch leaves running is wever_kill-switch-check. Verdict risk_blast_wide, risk_blast_partial, risk_blast_unmapped or risk_blast_contained. POST /api/risk-blast-radius-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-compounding-checkAgent Risk: submit one planned action as the series it runs in: the action with its id, what one run puts at stake, its unit, how many times it runs, the span the runs fall in, the…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,848 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
Description, as the server wrote it Agent Risk: submit one planned action as the series it runs in: the action with its id, what one run puts at stake, its unit, how many times it runs, the span the runs fall in, the total the plan states for the whole series, whether a fault in one run repeats in every run (such as one wrong address) and its description, read for wording that repeats without an end, such as until it succeeds, keep retrying or indefinitely; each follow-on it sets off with its id, name, what it puts at stake and whether it runs automatically, after an approval or on a condition (an empty list when none); the most the series and its follow-ons may put at stake, with its unit; and optionally the share of that limit that counts as elevated (80 when left out). The signed report gives the action with any unbounded wording, every follow-on with its trigger, the series with the stake per run, the runs, their total, the follow-ons’ total, the series total and its share of the limit, and counts the runs, follow-ons, automatic follow-ons and follow-ons with no stake, and names a runaway series: a series over its limit and a plan that repeats with no stated end; then a series near its limit, runs that fail together, a follow-on that runs by itself, a plan that states a smaller series total than its runs add up to and a plan that sizes only one run, with the fix for each. This reads one action as the series of runs and follow-ons it sets off, past the single-action view: a system’s retry plan is wever_retry-plan-check, where a payment retry can charge twice is wever_retry-safety-check, the effects a resumed or retried task can repeat are wever_duplicate-effect-check and duplicate-looking payments in a transaction list are wever_spending-anomaly-check. Verdict risk_compounding_runaway, risk_compounding_elevated, risk_compounding_untracked or risk_compounding_bounded. POST /api/risk-compounding-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-concentration-checkAgent Risk: submit one planned action and where value already sits: the action with its id, where it puts value (such as one supplier, one wallet or one provider), the kind of plac…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,877 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
Description, as the server wrote it Agent Risk: submit one planned action and where value already sits: the action with its id, where it puts value (such as one supplier, one wallet or one provider), the kind of place (a wallet, a vendor, a position, a provider, a counterparty or other), who stands behind that place when it differs (such as the custodian or the parent company), what it puts there, its unit and its description; each holding with its id, place, kind, owner, amount and unit (an empty list when nothing is held); and optionally the most one place or owner may hold as a share of the total (50 when left out) and the share that counts as elevated (25 when left out). The signed report gives the action, every place with what it holds before and after the action, its shares and its status as over_cap, elevated, spread or unread, the shares before and after for the action’s place and for its owner across every place it stands behind, and counts the places, the holdings counted and those of another kind or unit, and names an excessive concentration: the action’s place over the cap after the action, and its owner over the cap across its places; then a share at or above the elevated line, an action that adds to the place already holding the most and another place already over the cap, with the fix for each. Holdings of the action’s kind and unit are counted, places compared as plain words. This reads where one action puts value against the spread already held, beside the counterparty quality the wever_vendor checks read: stated concentration caps per asset or chain against current balances are wever_treasury-policy-check, a vendor’s lock-in and exit terms are wever_vendor-lockin-check and a counterparty’s stated identity against its evidence is wever_vendor-identity-check. Verdict risk_concentration_excessive, risk_concentration_elevated, risk_concentration_unknown or risk_concentration_spread. POST /api/risk-concentration-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-dependency-checkAgent Risk: submit one planned action and what it depends on: the action with its id, when it starts and completes and its description, read for assumptions it leans on without lis…4 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...a price, a person, an approval, data, a credential or other), whether the action fails or..."
- Info
The input schema at properties.dependencies.items.properties.kind.enum[5] mentions "credential".
What this check looks forMatched text
inputSchema.properties.dependencies.items.properties.kind.enum[5]: "credential"
- Info
The input schema at properties.dependencies.items.properties.kind.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.dependencies.items.properties.kind.description: "...a price, a person, an approval, data, a credential or other."
- Info
The description is 3,774 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
Description, as the server wrote it Agent Risk: submit one planned action and what it depends on: the action with its id, when it starts and completes and its description, read for assumptions it leans on without listing them, such as assuming, as long as or provided that; each dependency with its id, name, kind (a service, a price, a person, an approval, data, a credential or other), whether the action fails or goes wrong without it, the fallback, how many other sources could stand in for it, when it was last confirmed available or true and when it stops holding (such as a quote’s expiry or an approval’s end); optionally the moment to read at (the action start when left out); and the hours since confirmation that count as stale (24 when left out). The signed report gives the action with any unlisted assumptions, every dependency with its hours since confirmation and its status as covered, minor, single_point, lapses, stale, unrated or unconfirmed, and counts the dependencies, the critical ones, the single points of failure, those lapsing, those stale and those covered, and names a fragile action: a critical dependency with no fallback and no other source, and a dependency that stops holding before the action completes; then a stale confirmation and an assumption the plan leans on without listing it, with the fix for each. A dependency not rated or never confirmed leaves the action unmapped. This reads what one action relies on staying available or true: an operator’s key custody, successor and records export are wever_continuity-check, the inputs a multi-agent plan waits on are wever_coordination-plan-check and the assumptions under a delivered result are wever_uncertainty-assumption-check. Verdict risk_dependency_fragile, risk_dependency_partial, risk_dependency_unmapped or risk_dependency_sound. POST /api/risk-dependency-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-downside-checkAgent Risk: submit one planned action’s outcomes, what the principal can absorb and how the downside was put to them: the action with its id and description; each outcome with its …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,080 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
Description, as the server wrote it Agent Risk: submit one planned action’s outcomes, what the principal can absorb and how the downside was put to them: the action with its id and description; each outcome with its id, what happens, what it costs the principal, its unit, its likelihood (likely, possible, unlikely or remote, a remote outcome read as not credible), whether the loss can be got back (such as through a chargeback or insurance) and whether the plan names it as the worst credible outcome; the principal’s capacity with the most loss they can absorb, its unit and where that figure comes from; the statement to the principal as written, read for wording that denies the downside such as risk-free or guaranteed, or softens it such as minimal risk; and optionally the share of the capacity at which a loss counts as severe (50 when left out). The signed report gives every outcome with whether it is credible and its share of the capacity, the capacity with the worst credible outcome and its share, the statement with any denying or softening wording and whether it gives the worst loss as a figure, and counts the outcomes credible, remote, recoverable and over the capacity, and names an unacceptable downside: a worst credible loss over what the principal can absorb, a statement that denies the downside while a credible loss exists and a named worst outcome smaller than the largest credible loss; then a worst loss at a large share of the capacity, a worst loss that cannot be got back, softening wording, a statement without the worst loss as a figure and no outcome named as the worst, with the fix for each. Outcomes compare only in the capacity’s unit. This reads the worst credible outcome of one action against what the principal can absorb: the value handed over before a counterparty pays in one proposed agreement is wever_negotiation-risk-check, who pays when an agent errs under pre-purchase terms is wever_liability-terms-check and the range around an estimate is wever_uncertainty-range-check. Verdict risk_downside_unacceptable, risk_downside_severe, risk_downside_unstated or risk_downside_acceptable. POST /api/risk-downside-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-exposure-checkAgent Risk: submit one planned action and the limits it runs under: the action with its id, the unit it counts in, what it puts at stake if it goes as planned, the most it puts at …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,231 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
Description, as the server wrote it Agent Risk: submit one planned action and the limits it runs under: the action with its id, the unit it counts in, what it puts at stake if it goes as planned, the most it puts at stake if it goes wrong (the worst case, not the expected case), how that worst case was reached, whether the worst case has no ceiling, such as an open-ended liability, whether a person approved it at its worst case and its description, read for wording that sizes it on the expected case such as should be fine or on average; each limit with its id, name, kind (a cap on one action, a cap over a period, a total cap or an amount above which a person approves), amount, unit and how much of it is already used; and optionally the share of a limit’s room that counts as near it (80 when left out). The signed report gives the action with its stake and where the stake comes from (the worst case, or the expected case when that is all the plan states), every limit with the room it leaves, the stake’s share of that room and its status as within, approved, near, over, room_unknown, other_unit or unread, and counts the limits within, near, over, with unknown room and in another unit, and names an excessive exposure: a stake over a limit’s room, a stake above an approval threshold with no approval recorded and a worst case with no ceiling; then a stake using most of a limit’s room, a plan that states only the expected stake, wording that sizes the action on the expected case and a worst case with no stated basis, with the fix for each. A cap on one action is read against the whole stake and a period or total cap against the room left after what is used; a limit in another unit is not compared. This reads one action’s worst-case stake against the agent’s limits, the limits the wever_limit checks review: a request’s allow-or-deny preview against a budget policy is wever_agent-budget-guard, a cap design’s scope, window and breach rule is wever_exposure-cap-check, which limit governs each kind of job is wever_limit-scope-check, one cap shared by several agents is wever_shared-cap-check and an agent budget’s five controls are wever_budget-controls-check. Verdict risk_exposure_excessive, risk_exposure_elevated, risk_exposure_unknown or risk_exposure_bounded. POST /api/risk-exposure-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_risk-timing-checkAgent Risk: submit one planned action’s timing: the action with its id, when the agent plans to act, how long it takes to complete (such as 2 hours) and its description, read for w…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,222 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
Description, as the server wrote it Agent Risk: submit one planned action’s timing: the action with its id, when the agent plans to act, how long it takes to complete (such as 2 hours) and its description, read for wording that pushes to act now, such as act now, limited time or last chance; each deadline or closing window with its id, name, when it falls, whether missing it loses the outcome rather than delaying it, who set it (the principal, the counterparty, law, a system, the agent or other) and whether it was confirmed at its source (an empty list when none applies); each scheduled event around the action with its id, name, when it falls and whether it stops the action completing, such as a settlement cutoff or a maintenance window; how far the price or condition the action rests on can move while it runs, against how far it can move before the action no longer makes sense; what waiting would add and what it would lose; and optionally the hours of slack that count as tight (24 when left out). The signed report gives every deadline with its slack in hours and its status as met, tight, missed, missed_soft or unread, every event as blocks, inside, outside or unread, the timeline with when the action completes and the nearest deadline, and counts the deadlines met, tight and missed and the events inside the run, and names harmful timing: a hard deadline after the action completes, an event that stops the action falling while it runs and conditions able to move further than the plan tolerates; then thin slack before a deadline, a soft deadline missed, a counterparty’s deadline not confirmed at its source, wording that pushes to act now, movement at half the tolerance or more, another event while the action runs, and waiting that adds something while the deadlines leave room, with the fix for each. This reads whether acting now is itself the risk for one action: every date and obligation in a document is wever_deadline-check, whether open commitments fit their deadlines is wever_commitment-deadline-check, an item’s urgency on a stated scale is wever_triage-severity-check and the age of the information behind a result is wever_uncertainty-freshness-check. Verdict risk_timing_harmful, risk_timing_tight, risk_timing_unknown or risk_timing_sound. POST /api/risk-timing-check. Reads only the recorded material about the planned action in the call: no network request, fetch, money movement, message, configuration change or external write; the planned action is not taken, paused, approved or undone and no limit, position or dependency is changed. Agent Risk reads the consequences of an action the agent plans to take, before it takes it: how much it puts at stake against the agent’s limits, whether it can be undone, how far a failure would spread beyond its target, the worst credible outcome against what the principal can absorb, whether it piles value into one place, what it depends on staying available or true, whether acting now is itself the risk and whether repeating or chaining it multiplies the risk. It is not confidence in information (the wever_uncertainty checks read how sure the information behind a decision is; a risk check reads what the action costs if it goes wrong: uncertainty is about what the agent knows, risk is about what the action can do), not permission (wever_consent-scope-check reads whether the principal’s grant covers a use and wever_authority-scope-check reads whether the agent may act at all; a risk check reads the consequences of an action already permitted, and a permission never makes an action safe), not an agreement’s exposure (wever_negotiation-risk-check reads the value handed over before a counterparty pays under one proposed agreement; a risk check reads one action the agent is about to take) and not sorting work (the wever_triage checks sort, route and escalate a queue of work; a risk check assesses one planned action before it runs). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-definition-checkAgent Roles: submit an agent’s role description: the role with its id, title and version, the purpose it states, each task in scope with its id and the outcome it delivers, the tas…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,069 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
Description, as the server wrote it Agent Roles: submit an agent’s role description: the role with its id, title and version, the purpose it states, each task in scope with its id and the outcome it delivers, the tasks out of scope, each success measure with its target, the decisions the role may make alone and the description as written, read for open-ended wording such as other duties as assigned, as needed or help with anything. The signed report gives the purpose with whether it names a result, every task with its outcome and whether it is also excluded, the exclusions, every success measure with its target and any opinion wording, the open-ended wording found and the decisions, and counts them, and names a vague role: no purpose, no tasks in scope, no success measure, open-ended duty wording, a measure that leaves success to opinion such as looks good or happy, and a task listed both in and out of scope; then no exclusions, tasks with no outcome, measures with no target, a purpose that names an activity rather than a result and no decisions the role makes alone, with the fix for each and what would change the verdict. No description text leaves the definition unverified. This reads the role description an agent could act on without guessing: a goal’s outcome and success test are wever_goal-statement-check and the rule set an operator hands an agent is wever_adoption-readiness-check. Verdict role_definition_vague, role_definition_partial, role_definition_unverified or role_definition_clear. POST /api/role-definition-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-delegation-checkAgent Roles: submit the delegations of a role with what the written role lets be delegated: the role with its id, title, owner, the register where delegations must be recorded and …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,514 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
Description, as the server wrote it Agent Roles: submit the delegations of a role with what the written role lets be delegated: the role with its id, title, owner, the register where delegations must be recorded and each duty it names as delegable or kept back, with the most days and the largest amount and unit a delegate may hold; each delegation in effect with its id, the duty, the delegate, who made it, the amount and unit, when it starts and ends, who stays accountable and where it is recorded (the role register, a delegation record, a message or nowhere); and the date delegations are read at. The signed report gives the role’s delegation clause, every delegation with the clause it falls under, its length in days and whether it has lapsed, and counts the delegations within the role, those of a duty kept back, over the limit, with no end and unrecorded, and names unbounded delegation: a duty the role does not let be delegated, an amount past the role’s limit or no limit where the role sets one, no end date, a length past what the role allows, someone other than the role’s owner kept accountable and a delegation recorded only in a message or nowhere; then a record outside the register the role names, a delegation made by someone other than the owner, a lapsed delegation still listed and a limit in another unit, with the fix for each and what would change the verdict. No delegation clause, no owner, no delegations, or an unstated accountable person or record leaves the delegations unverified. This reads every delegation of one role against what the role allows: one delegation’s own written bounds are wever_delegation-scope-check, its expiry enforcement is wever_delegation-expiry-check, a system’s delegation records are wever_delegation-record-check and authority narrowing down a chain is wever_delegation-chain-check. Verdict role_delegation_unbounded, role_delegation_partial, role_delegation_unverified or role_delegation_bounded. POST /api/role-delegation-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-owner-checkAgent Roles: submit a role’s ownership with its records: the role with its id, title, when it last changed hands or was rewritten and its description as written, whose first line i…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,427 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
Description, as the server wrote it Agent Roles: submit a role’s ownership with its records: the role with its id, title, when it last changed hands or was rewritten and its description as written, whose first line is read for the owner; the owner with the name, kind (a person, a team, an agent or a placeholder), whether the description names them at the top, the decisions they hold over the role (scope, permissions, delegation, handoff and retirement), when they took the role, when they last confirmed it and whether they are still in post; the backup or successor with the name and kind; and the date ownership is read at. The signed report gives the owner with the kind read, placeholder names such as the team read as placeholders, the first line of the description, the decisions held and lacking, the backup with whether it is a different person, and the currency of ownership against the date the role last changed hands, and counts them, and names a missing owner: no owner, an owner who is a team, an agent or a placeholder, an owner not named at the top, an owner no longer in post, an owner holding no decision over the role’s scope or permissions and ownership carried over from before the role last changed hands; then no backup, a backup who is the owner or not a person, decisions the owner lacks and a confirmation over a year old, with the fix for each and what would change the verdict. An unstated kind, top line, authority, assignment date, change date or post status leaves ownership unverified. This reads one role’s ownership: a production agent’s named owner, responder and fallback are wever_incident-ownership-check, the accountable person for each outcome is wever_ownership-map-brief and whether a registry entry names an owner is wever_agent-registry-check. Verdict role_owner_missing, role_owner_partial, role_owner_unverified or role_owner_named. POST /api/role-owner-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-permission-match-checkAgent Roles: submit the permissions an agent holds with its written role: the role with its id, title, version and when this version took effect; each permission the role calls for…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,523 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
Description, as the server wrote it Agent Roles: submit the permissions an agent holds with its written role: the role with its id, title, version and when this version took effect; each permission the role calls for with its system, level (read, write, admin or owner) and why; each permission the agent holds now with its id, system, level, when it was granted, when it was last used and how it was granted (through the role, directly, inherited or copied); the date access is read at; and optionally the days without use after which a permission counts as unused (90 when left out). The signed report gives every held permission with the level the role calls for, its grant date, its days unused and its status as matched, excess, above_role, predates_role, below_role, unused, carried, unrated or unread when no needs are listed, every need as held, missing or under_held, and counts them, and names mismatched permissions: a permission on a system the written role does not call for, a level above the role, a permission the role calls for that the agent lacks and a grant made before this version of the role took effect; then a level below the role, a permission the role names left unused past the window and one inherited or copied rather than granted through the role, with the fix for each and what would change the verdict. No list of needs or holdings, an unstated level or last use, or no reading or effective date leaves the permissions unverified. This reads the permissions an agent holds now against its current written role, with use and grant dates: the permissions granted at joining against the written job are wever_onboarding-permissions-check, one action’s access against its authority grant is wever_authority-scope-check and a stored grant evaluated for one subject, action and resource is wever_agent-permission-ledger. Verdict role_permissions_mismatched, role_permissions_partial, role_permissions_unverified or role_permissions_matched. POST /api/role-permission-match-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-review-checkAgent Roles: submit a role’s review cadence with its records: the role with its id, title, version and current holder; the schedule with the cadence (monthly, quarterly, semiannual…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,991 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
Description, as the server wrote it Agent Roles: submit a role’s review cadence with its records: the role with its id, title, version and current holder; the schedule with the cadence (monthly, quarterly, semiannual, yearly or none), the next due date and the reviewer; each review on record with its id, date, reviewer and outcome (confirmed, changed or retired); each change to the role’s work with its id, date and what changed, such as a new task, system or tool; and the date the review is read at. The signed report gives the schedule with the due date worked out and the days overdue, every review, and every change with whether it came after the last review, and counts them, and names a stale review: no cadence and no due date, no review on record, a review past due at the reading date and changes to the role’s work made after the last review while it lapsed; then no named reviewer, review by the role’s own holder, changes since a review not yet due and reviews with no outcome, with the fix for each and what would change the verdict. An unstated schedule, due date, reading date, change record or holder leaves the review unverified. This reads whether one role is kept current after it is written: the review of a threshold set is wever_threshold-review-check and an onboarding record’s review date is wever_onboarding-record-check. Verdict role_review_stale, role_review_partial, role_review_unverified or role_review_current. POST /api/role-review-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_role-separation-checkAgent Roles: submit the roles one agent holds with its approvals: optionally the agent with its id and name; each role with its id, title and duties (initiate, approve, execute, re…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,221 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
Description, as the server wrote it Agent Roles: submit the roles one agent holds with its approvals: optionally the agent with its id and name; each role with its id, title and duties (initiate, approve, execute, record, reconcile, review, pay or custody); optionally the duty pairs that must sit apart (initiate and approve, execute and approve, execute and review, execute and reconcile, record and custody, and pay and approve when left out); each piece of work with who made it and who approved it; and each conflict declared with its mitigation. The signed report gives every role with its duties and whether it holds both a maker and a checker duty, the rules read and every conflicting pair held with the roles that hold it and whether it is declared, the self-approved work, and every declaration with whether the conflict is still held, and counts them, and names conflicted roles: a duty pair that must sit apart held across the agent’s roles with no declaration, a maker and a checker duty in one role and work approved by the agent that made it; then declared conflicts still held, declarations with no mitigation and declarations for conflicts no longer held, with the fix for each and what would change the verdict. Roles with no stated duties and no approval record leave separation unverified. This reads separation of duties across the roles one agent holds: approver bands and an agent approving itself under a spending arrangement are wever_approval-chain-check and who checks a delegate’s returned work is wever_delegation-result-check. Verdict role_separation_conflicted, role_separation_partial, role_separation_unverified or role_separation_clean. POST /api/role-separation-check. Reads the role records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Agent Roles reads a role’s own records and checks the role itself: what the job is, who owns it, whether the permissions the agent holds match the written role, where its scope ends, whether a change of holder rewrote the description, whether it is reviewed on schedule, whether duties stay apart across the roles one agent holds, and whether delegations stay inside what the role allows. Beside it, wever_onboarding-permissions-check and wever_onboarding-readiness-check read a new agent’s grants and written job at joining, while a role check reads the written role and its records at any time after; wever_authority-scope-check reads one action’s access against its authority grant, while the permission match check reads every permission held against the role; and the Agent Delegation checks such as wever_delegation-scope-check read one delegation on its own terms, while the delegation check reads every delegation against what the role allows. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_router-disclosure-checkAgent Control: submit a router or marketplace’s stated ranking and routing criteria and its described decision flow (each step with the factors it applies and what it does).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Control: submit a router or marketplace’s stated ranking and routing criteria and its described decision flow (each step with the factors it applies and what it does). The signed report says whether a buyer or seller can tell why one seller was chosen, names every factor the flow applies that the stated criteria do not disclose, classed as paid placement, self-preference, inventory pressure, margin or commission, or another factor, and each stated criterion the flow does not use. Verdict undisclosed_influence, criteria_mismatch or disclosed. POST /api/router-disclosure-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-control-contract. Human page: https://weverlabs.com/agent-control/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_run-record-checkAgent Readiness: submit one workflow run’s records: the detection or trigger record (kind, rule, deterministic, time), investigation notes, approval records (approver, decision, ac…1 finding
Why it is in this group
- The name "wever_run-record-check" says "run": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_run-record-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_run-record-check" says "run": a word for a change.
Description, as the server wrote it Agent Readiness: submit one workflow run’s records: the detection or trigger record (kind, rule, deterministic, time), investigation notes, approval records (approver, decision, actions covered) and the actions taken, each with ids, links and timestamps. The signed report says whether the run is auditable end to end, tracing each action through its investigation and approval to the trigger, and names every missing or unlinked record (no approval record, actions with no matching investigation, unknown references) and every timestamp that is missing, unreadable or out of order. Verdict unprovable, gapped or auditable. POST /api/run-record-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-readiness-contract. Human page: https://weverlabs.com/agent-readiness/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_sandbox-isolation-checkAgent Governance: submit a test or sandbox environment design for agents: the credentials it holds (where each is accepted and how it is made), the systems, tool platforms and endp…4 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...dbox environment design for agents: the credentials it holds (where each is accepted and ho..."
- Info
The input schema at properties.credentials.items.properties.valid_in.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.valid_in.description: "Where the credential is accepted."
- Info
The input schema at properties.credentials.items.properties.form.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.form.description: "How the credential is made."
- Info
The input schema at properties.credentials.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.credentials.description: "Credentials the sandbox holds or can read."
Description, as the server wrote it Agent Governance: submit a test or sandbox environment design for agents: the credentials it holds (where each is accepted and how it is made), the systems, tool platforms and endpoints the agents use (production, shared or sandbox, whether the sandbox can connect, internet exposure, what each accepts to let a caller in, the access it grants and its rate limit), the data it holds or can read, and its outbound default. The signed report names every path from the test environment into production systems or data, production asset by production asset: credentials accepted in production, predictable, default or shared credentials, production or shared systems that take a sandbox credential or none, production data the sandbox can read, unrestricted outbound reach and reachable systems with no rate limit, with the fix for each. A single route from the sandbox to a production effect is wever_escape-path-check and check order inside one payment flow is wever_firewall-order-check. Verdict production_reachable, isolation_unverified or isolated. POST /api/sandbox-isolation-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_schedule-createStanding jobs: register a recurring job for the paying wallet.Reads text others writeCan send data outNo findings
Why it is in this group
- The name "wever_schedule-create" says "schedule": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Standing jobs: register a recurring job for the paying wallet. Name one read-only Wever tool from the schedulable list (checks, briefs, lookups and watch evaluations such as wever_refund-eligibility-check, wever_market-data-check or wever_watch-status), its input (string values may use {{run_at}} and {{run_date}}), a cadence (hourly, daily, weekly, or interval with interval_seconds from 900 to 604800), an optional start_at, a callback_url and an expires_at up to 30 days ahead, with optional max_runs. The tool validates the input before billing. Returns the job id and the exact schedule accepted: anchor, interval, expiry, planned runs and the first run times. Each run POSTs a signed run report carrying the tool’s own signed report to callback_url. A tool that moves money, places an order or changes state is refused with tool_not_schedulable. One call covers every run until the job ends. POST /api/schedule-create. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_schedule-listStanding jobs: list the calling wallet’s standing jobs with state (pending_activation, active, completed, expired or cancelled), runs completed and failed, last result and next run…1 finding
Why it is in this group
- The name "wever_schedule-list" says "schedule": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_schedule-list" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_schedule-list" says "schedule": a word for a change.
Description, as the server wrote it Standing jobs: list the calling wallet’s standing jobs with state (pending_activation, active, completed, expired or cancelled), runs completed and failed, last result and next run, plus the wallet’s digests and routing rules. Optional state filter. POST /api/schedule-list. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_schedule-statusStanding jobs: one job owned by the calling wallet in detail: schedule, runs completed, succeeded, failed and skipped, the last result summary, recent failures with error codes, re…1 finding
Why it is in this group
- The name "wever_schedule-status" says "schedule": a word for a change.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
The server marks "wever_schedule-status" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_schedule-status" says "schedule": a word for a change.
Description, as the server wrote it Standing jobs: one job owned by the calling wallet in detail: schedule, runs completed, succeeded, failed and skipped, the last result summary, recent failures with error codes, recent runs with delivery outcomes, the next run, and one full signed run report (the latest, or run_index). Past run records stay readable after the job ends. POST /api/schedule-status. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_second-date-betPlace a YES or NO stake using free nonredeemable play credits with no cash value.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Place a YES or NO stake using free nonredeemable play credits with no cash value. Anyone with the market ID and a signed wallet may participate. Hard caps: 5 credits per bet and 25 credits per market. POST /api/second-date-bet. API access costs 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls per rolling 30 days. This access fee is separate from play-credit stakes. Each wallet begins with 25 free play credits. X-Wever-Wallet-Proof binds the exact JSON arguments and API path. Human consent and reports use https://weverlabs.com/second-dates/. Betting closes at the deadline or first human report. Missing reports return stakes seven days after the close. Keep request_id and exact arguments for retry; use a new request_id for current status. Read /api/second-dates-contract for schemas and public receipt verification key.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_second-date-market-openCreate a market linked to your completed staked first date.Reads text others writeNo findings
Why it is in this group
- The name "wever_second-date-market-open" says "open": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Create a market linked to your completed staked first date. Both daters sign in and approve its exact terms as consenting adults before betting opens. POST /api/second-date-market-open. API access costs 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls per rolling 30 days. This access fee is separate from play-credit stakes. Each wallet begins with 25 free play credits. X-Wever-Wallet-Proof binds the exact JSON arguments and API path. Human consent and reports use https://weverlabs.com/second-dates/. Betting closes at the deadline or first human report. Missing reports return stakes seven days after the close. Keep request_id and exact arguments for retry; use a new request_id for current status. Read /api/second-dates-contract for schemas and public receipt verification key.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_second-date-resolveRead current status, resolve matching human reports, submit an operator-signed decision for conflicting reports, or verify a signed receipt.Reads text others write1 finding
Why it is in this group
- The name "wever_second-date-resolve" says "resolve": a word for reading.
- The server marks it as not read-only (readOnlyHint: false).
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Warn
The server marks "wever_second-date-resolve" as not read-only (readOnlyHint: false), and its words say it reads.
What this check looks forMatched text
annotations: {"readOnlyHint":false,"destructiveHint":false,"idempotentHint":true,"openWorldHint":true}; The name "wever_second-date-resolve" says "resolve": a word for reading.
Description, as the server wrote it Read current status, resolve matching human reports, submit an operator-signed decision for conflicting reports, or verify a signed receipt. Both YES reports select YES; both NO reports select NO. Winners share the play-credit pool pro-rata. POST /api/second-date-resolve. API access costs 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls per rolling 30 days. This access fee is separate from play-credit stakes. Each wallet begins with 25 free play credits. X-Wever-Wallet-Proof binds the exact JSON arguments and API path. Human consent and reports use https://weverlabs.com/second-dates/. Betting closes at the deadline or first human report. Missing reports return stakes seven days after the close. Keep request_id and exact arguments for retry; use a new request_id for current status. Read /api/second-dates-contract for schemas and public receipt verification key.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_security-disclosure-checkDiligenceOps: submit a security or vulnerability disclosure policy, or a security.txt file.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a security or vulnerability disclosure policy, or a security.txt file. The signed report marks a contact method, response timelines, scope, safe-harbor language and a PGP key or other secure channel present, missing or ambiguous, with quotes: the reporting contacts, each stated timeline with its stage (acknowledgement, triage, fix or disclosure), the in-scope and out-of-scope hosts, any line that reserves legal action, and for a security.txt file its fields and whether its Expires date has passed. POST /api/security-disclosure-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-disclosure-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_settled-value-checkAgent Oversight: submit a payment-metrics summary (transaction count and settled value, such as "196M" and "$42M", optional asset, window, unique payers and payees, and a size dist…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit a payment-metrics summary (transaction count and settled value, such as "196M" and "$42M", optional asset, window, unique payers and payees, and a size distribution as buckets with count or value). The signed report gives the average ticket, the share of value in payments of $1 or more, churn-shaped patterns such as a very high count at a very low value, and the questions the numbers cannot answer. Verdict churn_shaped, distribution_needed or value_backed. POST /api/settled-value-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_settlement-proof-checkAgent Access: submit a settlement claim (transaction reference, amount, asset, destination, network, what the claim rests on, whether delivery was made against it), the chain evide…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Access: submit a settlement claim (transaction reference, amount, asset, destination, network, what the claim rests on, whether delivery was made against it), the chain evidence read for it, each reading with its source, whether that source stands outside the seller and its facilitator, whether it found the transaction, its status (success, reverted, pending or dropped), the amount, asset, destination, network, confirmations, finality and when it was read, and the confirmations the receiver requires. The signed report says whether the transaction exists, is final and matches the claim field by field, and names every claim whose only proof is the seller’s own receipt, record or facilitator response, every settlement that reverted or was dropped after an earlier success, and every delivery made before finality. Two records agreeing with each other are wever_settlement-reconciliation-check, an invoice against a payment record is wever_invoice-payment-match and proof of a code change is wever_change-proof-check. Verdict settlement_failed, self_attested, settlement_unverified or settlement_proven. POST /api/settlement-proof-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_settlement-reconciliation-checkCommerce Assurance: submit a list of expected settlements (reference, amount, currency) and a list of received transfers, with an optional default currency and tolerance.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Commerce Assurance: submit a list of expected settlements (reference, amount, currency) and a list of received transfers, with an optional default currency and tolerance. The signed report gives deterministic totals for each side and the net per currency, then each item tied to its reference: matched, missing, duplicate, short with the amount, overpaid with the amount, and unmatched transfers. Verdict reconciled or exceptions_found. POST /api/settlement-reconciliation-check. Reads only the material in the call: no page fetch, no account access, no counterparty contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/commerce-assurance-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_shipping-checkDiligenceOps: submit a shipping promise (as quoted text or as dispatch time, delivery window, delivery-by date, cost and free-shipping threshold) and the delivery evidence (order, …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a shipping promise (as quoted text or as dispatch time, delivery window, delivery-by date, cost and free-shipping threshold) and the delivery evidence (order, dispatch and delivery dates or tracking events, the shipping charge and order total). The signed report computes each deadline in calendar or business days and lists the dispatch, delivery and cost mismatches with the days late and the amount over. POST /api/shipping-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_single-credential-issuePaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.2 findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Findings
- Info
The title mentions "credential".
What this check looks forMatched text
title: "Issue a verified single credential"
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...d over rolling 30 days. Issue the exact credential authorized by a human at https://weverl..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Issue the exact credential authorized by a human at https://weverlabs.com/verified-single/ after existing Wever email sign-in, a current photo and their own single-status attestation. Submit their short-lived issuance_authorization. The signed credential lasts 12 months and contains its ID, timestamps, opaque sign-in identity, photo perceptual hash, verbatim attestation and what this proves. Identity means verified control of the email account. Single status is self-attested, not independently investigated by Wever. POST /api/single-credential-issue. 0.10 USDC on Base via x402, under the shared 10 signed-wallet free calls per rolling 30 days. Replay the same authorization for the original result. Check status anonymously with wever_single-credential-check on /mcp-free.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_sla-credit-checkDiligenceOps: submit SLA terms and an outage or service report, as text or as the month, outages, total downtime and stated uptime, with an optional monthly fee.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit SLA terms and an outage or service report, as text or as the month, outages, total downtime and stated uptime, with an optional monthly fee. The signed report works out the uptime exactly from the period minutes and the counted downtime (setting scheduled maintenance or degraded time apart when the terms exclude it), says whether the commitment was missed and by how many points, and names the credit band the terms prescribe with the credit percentage, any cap, the credit amount and the credit request deadline, each quoted. POST /api/sla-credit-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_sla-evidence-checkDiligenceOps: submit an SLA or service terms text.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit an SLA or service terms text. The signed report states the uptime commitment figure and its period, the service-credit terms and bands, the exclusions and the notice requirements with their deadlines, each present or missing and quoted briefly when present, and flags a commitment stated with two figures. POST /api/sla-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_spec-drift-checkAgent Oversight: submit spec statements and notes on what the implementation does (each with optional id, subject key and where).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Oversight: submit spec statements and notes on what the implementation does (each with optional id, subject key and where). The signed report names every point where behavior conflicts with the spec, each classified by consequence (money-moving first, then state-changing, then cosmetic) with the conflicting figures or wording, the statements no note addresses or whose figures do not compare, and behavior the spec does not mention. Use it to report a conflict as a finding before shipping. Verdict money_moving_conflict, state_changing_conflict, cosmetic_conflict, unresolved or consistent. POST /api/spec-drift-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-oversight-contract. Human page: https://weverlabs.com/agent-oversight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_spend-category-briefAgent Treasury: submit one period of transactions (amount, asset, counterparty, category where given, timestamp).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Treasury: submit one period of transactions (amount, asset, counterparty, category where given, timestamp). The signed brief gives totals and shares by category, the top counterparties by spend, the largest single payments and how many transactions arrived uncategorized. Any period; the month-end close pack with receipts, refunds and exceptions is wever_expense-close-pack. Verdict fully_categorized or partly_uncategorized. POST /api/spend-category-brief. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_spending-anomaly-checkAgent Treasury: submit a list of transactions (amount, asset, counterparty, timestamp, category where given) with your stated budget rules: a per-transaction cap, per-counterparty …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Treasury: submit a list of transactions (amount, asset, counterparty, timestamp, category where given) with your stated budget rules: a per-transaction cap, per-counterparty or per-category caps per day, week, month or in total, and allowed counterparties or categories. The signed report flags every transaction that breaks a stated rule, every duplicate-looking payment (one counterparty, one amount, close timestamps) and every amount that is an outlier against the supplied set itself, each flag naming the rule or pattern and the transactions involved. Verdict rules_broken, patterns_found, rules_unconfirmed or no_flags. Live wallet activity is wever_wallet-statement. POST /api/spending-anomaly-check. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_spending-policy-checkAgent preflight round 3: check your own agent spending policy before the agent runs under it.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight round 3: check your own agent spending policy before the agent runs under it. Input text, fields (per_call_cap, daily_cap, allowed_assets, allowed_networks, recipient_allowlist, approval_thresholds, kill_switch, categories with optional caps, currency) or both. The signed report marks each control present, missing or ambiguous, with quotes, and names the dangerous shapes plainly: a category with no cap, a daily cap with no per-call cap, a per-call cap with nothing capping the total, spending stated as unlimited, any recipient, asset or network allowed, and no way to stop the agent. POST /api/spending-policy-check. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_sports-plan-tripAssemble round-trip flights, hotel nights, event tickets and a dinner table for a specific game or event in any sport, using the home city, party size and budget cap.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Assemble round-trip flights, hotel nights, event tickets and a dinner table for a specific game or event in any sport, using the home city, party size and budget cap. Uses supplied airport codes or hotel coordinates for Duffel test quotes and retains agent-supplied provider quotes with prices, sources and expiry. Marks remaining legs as quote needed. Returns an itemized itinerary, quoted subtotal, budget targets and a principal review link. POST /api/sports-plan-trip. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. API access fees are separate from travel spending. Per-leg principal approval and delegated authority govern spending. Conditional escrow has a 300 USDC per-leg cap. Duffel test mode uses a test token and simulated travel. Booking approval is separate from quote selection. Uncertain attempts are reconciled through trip-status. Read /api/sports-trips-contract for schemas, provider availability, receipt verification and approval terms.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_state-reconciliation-checkAgent Recovery: submit a recovery flow: each piece of state the agent works from (a balance, task states, sent items, orders, bookings, records or something else) with the system r…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,045 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/."
Description, as the server wrote it Agent Recovery: submit a recovery flow: each piece of state the agent works from (a balance, task states, sent items, orders, bookings, records or something else) with the system record it is checked against, when the agent’s view is checked against it after an interruption (before new work, after new work has started, on a schedule or never) and what happens when they differ (the system record wins, the agent’s view wins, work stops for a person or the difference is only logged), and whether new work waits for reconciliation to finish, plus optionally the recoveries seen with when reconciliation finished, when new work started and the differences found and settled. The signed report says whether the agent’s view is reconciled against the system’s record before new work starts, and names every view never checked, checked only after new work or trusted over the record, no hold on new work and every recovery seen starting work before reconciliation (the agent resuming on its own stale picture), then every check only on a schedule, every difference only logged and every difference left unsettled, with the fix for each. This reads agreement on state after an interruption: expected settlements against received transfers are wever_settlement-reconciliation-check, cached context that can outlive a change is wever_cache-staleness-check and what an interrupted task’s reported state shows is wever_partial-state-check. Verdict resumes_on_stale_state, reconciliation_gaps, reconciliation_unverified or reconciled_before_work. POST /api/state-reconciliation-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_submit-workThe accepted worker submits a result and evidence for the human to review.Reads text others writeNo findings
Why it is in this group
- The name "wever_submit-work" says "submit": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it The accepted worker submits a result and evidence for the human to review. Binds the submission digest to this job and its acceptance criteria. POST /api/submit-work. 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. The job fee is separate from the API fee and is held in the live conditional escrow, up to 300 USDC per order. Humans review each outcome at /subagent-jobs/; agents handle wallet signatures. Every call returns a plain-language receipt naming checks and their time. Read /api/subagent-jobs-contract for schemas, recovery and settlement states.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_subprocessor-evidence-checkDiligenceOps: submit subprocessor claims, as a statement, a list or entries with name, purpose and location, and the agreement or subprocessor list offered for them.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit subprocessor claims, as a statement, a list or entries with name, purpose and location, and the agreement or subprocessor list offered for them. The signed report pairs the processors by name (legal suffixes set aside, AWS and Amazon Web Services as one) and lists those missing from the evidence, those the evidence adds and those whose purpose or location changed, with quotes. POST /api/subprocessor-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_subscription-auditDiligenceOps: submit up to 200 subscriptions or recurring charges (name, amount, currency, cadence, last charge date, prior amounts and charge dates when known).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit up to 200 subscriptions or recurring charges (name, amount, currency, cadence, last charge date, prior amounts and charge dates when known). The signed report flags duplicate services, price increases against prior amounts, charge intervals off their cadence and double charges, and charges whose last date is stale, with annual and monthly equivalents per currency. POST /api/subscription-audit. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_succession-briefAgent preflight round 2: a short brief of an operator’s succession arrangements.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent preflight round 2: a short brief of an operator’s succession arrangements. Input text, fields (successor, trigger, key_transfer, account_transfer, timeline, receives) or both. The signed brief restates the arrangement in order: who takes over, what sets it off, how the keys and the accounts pass, on what timeline and what the successor receives, in the arrangement’s own words, and names each gap plainly: no successor named, no trigger defined, no transfer path for keys or accounts, no timeline, or nothing said about what the successor receives. POST /api/succession-brief. Reads only the material in the call: no page fetch, no registry or chain lookup, no counterparty contact, and no payment signed, submitted or moved. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-preflight-contract. Human page: https://weverlabs.com/preflight/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_support-briefDiligenceOps: submit one support conversation excerpt, as a chat, email thread or ticket.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit one support conversation excerpt, as a chat, email thread or ticket. The signed brief states who speaks for support and who is the customer, the issue raised, the current status, every promise support made with its amount and deadline, and the stated next step with who owns it, each quoted. POST /api/support-brief. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-vendor-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_terms-change-checkJob Records: submit two versions of the same terms, policy or agreement text (the version in force at purchase or signing and the current version, each with its effective date wher…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Job Records: submit two versions of the same terms, policy or agreement text (the version in force at purchase or signing and the current version, each with its effective date where stated) and the comparison date. The signed report splits both into sections, aligns them and lists every changed, added and removed section with a plain buyer-impact label: price or fee, refund or cancellation, liability or dispute, term or renewal, or removed, with a single changed figure named before and after. Two-version comparison: one document’s evidence is wever_terms-evidence-check and contract redlining is wever_contract-diff-check. Verdict buyer_impact_changes, other_changes_only or no_changes. POST /api/terms-change-check. Reads only the material in the call: no network request, fetch, wallet or merchant lookup, order, payment or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/job-records-contract. Human page: https://weverlabs.com/job-records/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_terms-evidence-checkDiligenceOps: read a terms or policy page (a public HTTPS url, read once with GET) or supplied text, up to 512 KB, and return a signed report on whether fee, refund, cancellation a…Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it DiligenceOps: read a terms or policy page (a public HTTPS url, read once with GET) or supplied text, up to 512 KB, and return a signed report on whether fee, refund, cancellation and ownership clauses are present or missing, with match counts, heading lines and up to three brief quotes per clause with their offsets. POST /api/terms-evidence-check. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-contract. Human page: https://weverlabs.com/diligence/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_terms-match-checkAgent Governance: submit purchases, each with its quoted or recorded terms and the governing published or contract terms (price, scope and cancellation, with where each came from) …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Governance: submit purchases, each with its quoted or recorded terms and the governing published or contract terms (price, scope and cancellation, with where each came from) and whether the two were compared before payment. The signed report names every mismatch between the terms the purchase was made on and the governing terms, with a price difference in percent of the governing price, and every purchase paid with no terms comparison before payment. One service’s terms are wever_service-terms-check, terms changing over time are wever_terms-change-check and settled money against invoices is wever_settlement-reconciliation-check. Verdict terms_mismatch, compared_late, match_unverified or terms_match. POST /api/terms-match-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-governance-contract. Human page: https://weverlabs.com/agent-governance/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_testimonial-evidence-checkDiligenceOps: submit attributed quotes or endorsements and the source material.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit attributed quotes or endorsements and the source material. The signed report marks each attribution as matching, differing or without a source, comparing the wording word by word (verbatim, with elisions, or a close version with the differing words) and checking that the attributed person, role and organization appear with the passage. POST /api/testimonial-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_threshold-review-checkAgent Limits: submit a threshold set as of a date, each threshold with its value, when it was set, who owns its review, its review cadence (such as quarterly, 90 days or none) and …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Limits: submit a threshold set as of a date, each threshold with its value, when it was set, who owns its review, its review cadence (such as quarterly, 90 days or none) and its review history (each review with its date, who held it, what it decided, the data it looked at and its rationale), plus the changes to the agent or its work since thresholds were set (a model upgrade, a new merchant, a new task). The signed report names every threshold never reviewed since setup, on no schedule, past its scheduled review, with no named owner, whose last review left no decision, data or rationale, and that the agent or its work has changed around since its last review, with the fix for each. This reads whether thresholds are kept current after setup: whether escalation triggers are operable is wever_escalation-threshold-check and whether consequential actions are gated by a threshold is wever_action-threshold-check. Verdict thresholds_fossilized, review_gaps, review_unverified or reviewed_on_schedule. POST /api/threshold-review-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-limits-contract. Human page: https://weverlabs.com/agent-limits/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-backing-checkTokenized Asset Purchase: submit a token’s claimed backing with its records: the claim with the backing claimed (full, over, partial or none), the reserve ratio claimed, the underl…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,497 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
Description, as the server wrote it Tokenized Asset Purchase: submit a token’s claimed backing with its records: the claim with the backing claimed (full, over, partial or none), the reserve ratio claimed, the underlying asset with its name, identifier (such as an ISIN, a CUSIP or a bar list) and custodian, and the wording, read for fully backed, 1:1, 100% reserves or audited; the supply with the tokens outstanding, the value each token claims and its date; each reserve, audit, attestation, proof-of-reserves or self-reported report with its id, date, author, the reserves it states and the tokens outstanding it states; each mint and burn record with its id, tokens, date and whether a matching deposit or payout is recorded; the date the backing is read at; and optionally the days a report may age (45 when left out). The signed report gives the claim with the ratio claimed, the underlying and the wording found, the reserves with every report, the latest report’s age, the tokens outstanding, the value each claims and the reserve ratio computed, and the ledger with the tokens minted and burned, the net beside the supply and the mints with no deposit, and counts them, and names unproven backing: an underlying asset not identified, reserves below the ratio claimed, a latest report past its allowed age, mint and burn records that do not net to the supply, mints with no matching deposit and full-backing or audit wording the records contradict; then a self-reported latest report and a report supply that differs from the supply record, with the fix for each and what would change the verdict. An unstated claim, no reports, reserves, supply or unit value, no reading date or report date, no mint and burn records and an underlying named with no identifier leave the backing unverified. This reads one token’s backing against its own records: the NAV the Atlas dataset records is wever_atlas-nav-evidence and a claim against offered evidence is wever_claim-evidence-check. Verdict tokenized_backing_unproven, tokenized_backing_partial, tokenized_backing_unverified or tokenized_backing_proven. POST /api/tokenized-backing-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-custody-checkTokenized Asset Purchase: submit the custody arrangement for a purchased token: who holds the keys (the buyer, a custodian, the platform, the issuer or several parties); segregated…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,470 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
Description, as the server wrote it Tokenized Asset Purchase: submit the custody arrangement for a purchased token: who holds the keys (the buyer, a custodian, the platform, the issuer or several parties); segregated or omnibus holding; the custodian with its name, regulator, license, status (active, suspended or revoked) and when the status was checked; how client assets stand if the custodian fails, whether bankruptcy remote and whether outside the custodian’s estate; the failure path with a successor custodian, the recovery process and the days to restore access; the latest attestation with its date, its author and whether it is independent; the custody wording, read for segregated, bankruptcy remote, your keys or self-custody; and the date custody is read at. The signed report gives the keys with whether a third party holds them and the wording found, the custodian with its regulator, license, status, the days since the status check and the attestation with its age, the segregation with the insolvency protection, and the failure path, and counts them, and names opaque custody: keys held by a third party with no custodian named, a custodian whose status is suspended or revoked, client assets inside the custodian’s estate, omnibus holding with no protection if the custodian fails and segregation, self-custody or bankruptcy wording the records contradict; then protected omnibus holding, an attestation over a year old or not independent, a status check over 90 days old and a recovery path with no time to restore access, with the fix for each and what would change the verdict. Unstated keys, holding, custodian status, regulator, insolvency treatment, failure path, attestation and reading date leave custody unverified. This reads one token’s custody records: the custody and backing field the Atlas dataset records per issuer is wever_atlas-issuer-brief and one task passed between agents step by step is wever_chain-of-custody-check. Verdict tokenized_custody_opaque, tokenized_custody_partial, tokenized_custody_unverified or tokenized_custody_sound. POST /api/tokenized-custody-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-platform-terms-checkTokenized Asset Purchase: submit the issuing or trading platform’s terms with the buyer and the purchase: optionally the platform with its id and name; the buyer with the jurisdict…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,774 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
Description, as the server wrote it Tokenized Asset Purchase: submit the issuing or trading platform’s terms with the buyer and the purchase: optionally the platform with its id and name; the buyer with the jurisdiction, investor type (retail, accredited, qualified or institutional), whether identity checks are complete, the days the buyer means to hold and the purchase amount; the terms with the investor types the platform admits, the jurisdictions it admits and excludes, whether identity checks are required, the minimum and maximum purchase, the lockup in days, the platform’s right to freeze holdings and its right to claw them back, each with the conditions written for it, the fee schedule as written with each fee’s label, amount or percent, and the terms’ wording, read for sole discretion, may freeze, claw back, reverse or suspend without notice; and each fee charged on this purchase with its label and amount. The signed report gives the eligibility with the buyer’s jurisdiction and investor type against the lists, the identity checks and the purchase limits, the lockup beside the planned hold, the freeze and clawback rights with their conditions and the wording found, and every charge as matches, differs, not_in_schedule or unread against the written schedule, and counts them, and names blocking terms: an excluded jurisdiction or investor type, identity checks not complete, a purchase outside the limits, a lockup past the planned hold, a freeze or clawback right with no written conditions and a charge the schedule does not hold or sets at another amount; then freeze or clawback rights under written conditions and discretionary wording with no right set out, with the fix for each and what would change the verdict. No eligibility lists, an unstated buyer jurisdiction, investor type or identity status where the terms limit them, no lockup, unstated freeze or clawback rights and no written fee schedule leave the terms unverified. This reads one platform’s terms against one purchase: fee, refund and ownership clauses found in a terms page are wever_terms-evidence-check, a paid service’s terms from the paying side are wever_service-terms-check and liability terms agreed before an agent buys are wever_liability-terms-check. Verdict tokenized_terms_blocking, tokenized_terms_partial, tokenized_terms_unverified or tokenized_terms_acceptable. POST /api/tokenized-platform-terms-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-quote-checkTokenized Asset Purchase: submit a purchase quote for a tokenized asset with its stated terms: optionally the asset with its id, name, ticker and chain; the quote with its id, the …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,622 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
Description, as the server wrote it Tokenized Asset Purchase: submit a purchase quote for a tokenized asset with its stated terms: optionally the asset with its id, name, ticker and chain; the quote with its id, the price per token, the currency, the tokens the buyer receives, the total, each fee with its label and amount, the spread it states, the NAV or mid price it sits against, its minimum, when it was quoted and when it lapses, what the buyer receives (the token itself, a wrapped token, a claim on the platform, a fund unit or other) and its wording, read for no fees, zero spread, at NAV or price guaranteed; the terms with the price basis (NAV, mid, fixed or market), the most spread they allow, the minimum, the currency, the minutes a quote must stay open and what they promise the buyer receives; and the order with the amount the buyer means to spend and when the buyer would accept. The signed report gives the quote with its fees and the total stated beside the total recomputed from the tokens, the price and every fee, the spread stated beside the spread recomputed from the reference price and the most the terms allow, the expiry with the minutes open and whether it lapses before acceptance, and what the buyer receives beside the terms, and counts them, and names a misleading quote: a total that does not add up, a spread above the one stated or the one allowed, a quote that lapses before acceptance, an order under the minimum, a different instrument delivered and fee-free, spread-free or price-locked wording the figures contradict; then a spread the quote leaves unstated and a minimum, currency or validity that differs from the terms, with the fix for each and what would change the verdict. No tokens or total, no reference price or spread, no expiry or acceptance time, no instrument and no terms leave the quote unverified. This reads one token purchase quote against its own terms: conversion quotes ranked by effective rate are wever_exchange-quote-check, provider quotes for one job compared are wever_quote-compare and the NAV the Atlas dataset records is wever_atlas-nav-evidence. Verdict tokenized_quote_misleading, tokenized_quote_partial, tokenized_quote_unverified or tokenized_quote_clear. POST /api/tokenized-quote-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-redemption-checkTokenized Asset Purchase: submit a token’s redemption terms as written with the buyer: the rights with whether the token redeems, what it redeems into (the underlying asset, cash, …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,337 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
Description, as the server wrote it Tokenized Asset Purchase: submit a token’s redemption terms as written with the buyer: the rights with whether the token redeems, what it redeems into (the underlying asset, cash, a stablecoin or none), who may redeem (any holder, whitelisted holders, institutional holders or at the issuer’s discretion), the minimum in tokens, the notice and settlement days, the redemption fee, each gate, suspension, queue or cap the issuer may impose with its condition and limit, and the wording, read for redeem at any time, instant redemption, guaranteed, sole discretion or may suspend; the holder with the investor type, whether whitelisted, the tokens the buyer will hold and the days within which the buyer may need cash back; and past redemptions on record with each one’s id, date, status (paid, delayed, refused or suspended) and days. The signed report gives the rights with who may redeem against the buyer and the minimum against the holding, the timing with notice plus settlement days against the days the buyer may need, every gate with its condition, and the history, and counts them, and names a blocked redemption: no redemption right, redemption only at the issuer’s discretion, a buyer the terms do not let redeem, a holding under the minimum, a path slower than the buyer needs, refused or suspended redemptions on record and any-time wording the terms contradict; then gates and suspensions on the terms, delays on record and a redemption fee, with the fix for each and what would change the verdict. Unstated rights, redemption target, redeemers, timing, minimum and the buyer’s type, holding or need leave the path unverified. This reads one token’s redemption terms against one buyer: rights fields the Atlas dataset marks not disclosed are wever_atlas-rights-gap. Verdict tokenized_redemption_blocked, tokenized_redemption_partial, tokenized_redemption_unverified or tokenized_redemption_open. POST /api/tokenized-redemption-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-settlement-checkTokenized Asset Purchase: submit a tokenized asset purchase’s settlement terms with the trade and its settlement records: the settlement with how delivery and payment are tied (ato…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,532 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
Description, as the server wrote it Tokenized Asset Purchase: submit a tokenized asset purchase’s settlement terms with the trade and its settlement records: the settlement with how delivery and payment are tied (atomic, escrow, payment first or delivery first), the chain the token is delivered on, the payment chain and asset, the delivery and payment deadlines, where ownership becomes final (on chain, with a transfer agent or on the platform’s ledger), what happens if settlement fails with whether the buyer is refunded, in how many days and on what plan, and the wording, read for atomic, instant settlement, T+0 or delivery versus payment; the trade with the chains the buyer pays and receives on, the payment asset and when it traded; each settlement record with its id, leg (payment or delivery), status (settled, pending or failed), time and chain; and the time settlement is read at. The signed report gives the legs with the settlement chains and asset beside the buyer’s, the finality and the wording found, the timing with the delivery window, every record with whether a pending delivery is past its deadline, and the failure terms, and counts them, and names uncertain settlement: payment before delivery with no refund, no refund if settlement fails, a chain or asset the buyer does not pay or receive on, a payment settled with its delivery failed or overdue and atomic or instant wording the terms contradict; then ownership final off chain, legs still pending and a refund with no stated time, with the fix for each and what would change the verdict. An unstated model, delivery chain, finality, failure path, delivery deadline or buyer chain leaves settlement unverified. This reads one purchase’s settlement terms and records: a settlement claim against chain evidence is wever_settlement-proof-check, an agreed deal’s payment and delivery order is wever_negotiation-settlement-check and expected settlements matched to received transfers are wever_settlement-reconciliation-check. Verdict tokenized_settlement_uncertain, tokenized_settlement_partial, tokenized_settlement_unverified or tokenized_settlement_certain. POST /api/tokenized-settlement-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tokenized-transferability-checkTokenized Asset Purchase: submit a token’s transfer restrictions with its secondary venues and the buyer: the token with whether holders must be whitelisted, the date its lockup en…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,229 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
Description, as the server wrote it Tokenized Asset Purchase: submit a token’s transfer restrictions with its secondary venues and the buyer: the token with whether holders must be whitelisted, the date its lockup ends, who must approve a transfer (no one, the issuer, a transfer agent or the platform), the transfer fee, the jurisdictions barred from holding it and its wording, read for freely transferable, no restrictions or permissionless; each secondary venue with its name, kind (an exchange, an ATS, a DEX, OTC or the platform), whether it is open to the buyer and when it last traded the token; the holder with whether the buyer and the counterparty are whitelisted, the buyer’s jurisdiction and when the buyer may need to transfer; and the date transfers are read at. The signed report gives the restrictions with the wording found, every venue as trading, open, stale, closed or unread with the days since its last trade, and the holder, and counts them, and names a locked token: a lockup that runs past when the buyer may need to transfer, a whitelist the buyer or the counterparty is not on, a barred jurisdiction, no venue open to the buyer and free-transfer wording the restrictions contradict; then a transfer approval, a transfer fee, open venues with no trade in 30 days and a lockup that ends in time, with the fix for each and what would change the verdict. An unstated whitelist rule, approval rule, venue list, transfer date, whitelisting, jurisdiction or reading date leaves transfers unverified. This reads one token’s transfer restrictions after purchase: two Atlas assets’ rights side by side are wever_atlas-compare and a consent grant’s onward rule is wever_consent-transfer-check. Verdict tokenized_transfer_locked, tokenized_transfer_partial, tokenized_transfer_unverified or tokenized_transfer_free. POST /api/tokenized-transferability-check. Reads the purchase records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment or external write. Tokenized Asset Purchase reads the buyer’s own records for one purchase of a tokenized asset and checks each step before money moves: the quote against its stated terms, the platform’s terms against the buyer, the custody arrangement, the redemption path, the transfer restrictions, the settlement, the claimed backing and the full cost. Beside it, wever_atlas-asset-passport and wever_atlas-redemption-check read what the published Tokenization Atlas dataset records about an asset by its id, while a purchase check reads the records the buyer supplies for one purchase; and wever_vendor-quote-check and wever_price-truth-check read quotes and charges for goods, services and paid calls, while the quote and cost checks read a token purchase against its own terms. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/tokenized-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_tool-exposure-checkAgent Access: submit an API inventory: every endpoint or operation with what it can do (read, create, update, delete, admin, move money, personal data, send, execute or export), th…Can send data outNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a URL (input.endpoints): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Agent Access: submit an API inventory: every endpoint or operation with what it can do (read, create, update, delete, admin, move money, personal data, send, execute or export), the most sensitive data it handles, whether agents can call it as a tool and how it became one (curated, generated from an API description, passed through by a gateway or inherited), and the decision on exposing it with its outcome, its named decider and its date. The signed report names every endpoint that can delete, administer, move money or reach personal or confidential data and is exposed to agents with no recorded decision, exposed by generation or passthrough alone, exposed by a decision that names no decider, or exposed although its decision withholds it, with the fix for each. Spending caps are wever_exposure-cap-check, an agent’s access against its authority is wever_authority-scope-check and a client’s MCP configuration is wever_mcp-server-check. Verdict undecided_exposure, decision_gaps, exposure_unverified or decided_exposure. POST /api/tool-exposure-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_trace-outcome-briefAgent Operations: submit one job’s execution trace: its steps with their tool calls, recorded failures and recoveries, which steps are required and which replace others, and the fi…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Operations: submit one job’s execution trace: its steps with their tool calls, recorded failures and recoveries, which steps are required and which replace others, and the final outcome as recorded. The signed brief classifies every failure as recovered (the job absorbed it), worked around (a different path finished the step) or fatal to its step, ties the classification to the recorded outcome, and names every failure and outcome the trace leaves unexplained. Trace analysis: per-job money is wever_job-cost-brief. Verdict outcome_unexplained, fatal_failures, failures_absorbed or no_failures. POST /api/trace-outcome-brief. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-operations-contract. Human page: https://weverlabs.com/agent-operations/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_transaction-gate-checkAgent Access: submit an agent-facing surface’s actions, each with whether it reads, writes or spends, the steps it passes through in order (prepare, approve and submit), whether it…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Access: submit an agent-facing surface’s actions, each with whether it reads, writes or spends, the steps it passes through in order (prepare, approve and submit), whether it can also be invoked directly, who approves it (a person, an agent, a policy or no one), how the submitted action is bound to the approved one (digest, reference or none), whether each step is audited and what a caller needs to reach it. The signed report names every write or spend that can skip prepare, human approval or submit, runs its steps out of order or has a direct path around them, every approval not made by a person or not bound to what is submitted, every unaudited or unauthenticated write, and every read wrongly forced through the gate. Approver bands are wever_approval-chain-check, gates by action class are wever_action-threshold-check and whether the executed call matches the approved one is wever_approval-record-check. Verdict gate_skippable, gate_gaps, gate_unverified or gated. POST /api/transaction-gate-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_translation-evidence-checkDiligenceOps: submit the same passage in two languages, with optional language tags.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit the same passage in two languages, with optional language tags. The signed report lists the figures, dates, names and obligation words (must, must not, may) that differ between the versions, each quoted with its paired passage in the other version. POST /api/translation-evidence-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-ancillary-checkAgent Travel: submit the extras on a booking with what was authorized and needed: optionally the authorization with the most the extras may cost in all, its currency and the kinds …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,484 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
Description, as the server wrote it Agent Travel: submit the extras on a booking with what was authorized and needed: optionally the authorization with the most the extras may cost in all, its currency and the kinds approved (seat, bag, meal, wifi, lounge, priority, insurance, upgrade or other); what the fare includes, with its checked bags and whether seat selection comes with it; what the traveler needs, with the checked bags and whether a chosen seat is needed; and each extra with its id, kind, segment, description, quantity, amount, currency and whether the ticket or its documents show it. The signed report gives the authorization, what the fare includes, what is needed, every extra with whether its kind is authorized and whether it repeats another on the same segment, and the totals with the amount left under the cap, the checked bags bought and the total by kind, and counts them, and names excess extras: an extra of a kind the authorization leaves out, the same seat, meal or other single extra bought twice for one segment, a seat or bags bought where the fare includes them, an extra charged and missing from the ticket and extras over the authorized cap; then an extra not itemized with a kind and amount, bags or seats beyond what the traveler needs and extras in another currency, with the fix for each and what would change the verdict. No authorization, no cap, no authorized kinds, no record of which charged extras the ticket shows or no statement of need leaves the extras unverified. This reads the extras on one booking against what was authorized and needed: a paid call settled more than once is wever_double-charge-check, add-ons on a pricing page against its terms are wever_pricing-tier-check and a quote’s priced lines are wever_vendor-quote-check. Verdict travel_ancillary_excess, travel_ancillary_partial, travel_ancillary_unverified or travel_ancillary_as_authorized. POST /api/travel-ancillary-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-authorization-checkAgent Travel: submit one travel booking with the travel authorization behind it: optionally the traveler with an id and name; the authorization with its id, the approver, when it w…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,618 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
Description, as the server wrote it Agent Travel: submit one travel booking with the travel authorization behind it: optionally the traveler with an id and name; the authorization with its id, the approver, when it was approved, its currency, the most the booking may cost, the first and last travel dates approved, the cabins or classes approved (economy, premium economy, business or first), its purpose and the authorization as written; and the booking with its reference, when it was made, who booked it, its currency, what it cost in all and each flight, rail, hotel, car or other segment with its id, kind, from and to, departure and arrival with their offsets, carrier, number, the cabin booked and its price. The signed report gives the authorization, the booking, the booked total against the cap with its share and any amount over, every segment’s dates read by day against the approved window, every flight or rail cabin against the highest cabin approved, and the approval with whether it came before the booking and whether the approver is the traveler or the booking agent, and counts them, and names a booking over its authorization: a total over the cap, a segment outside the approved dates, a cabin above the approved cabins and a booking made before the approval; then a total at 95% or more of the cap, a booking in another currency and an approver who is the traveler or the booking agent, with the fix for each and what would change the verdict. No cap, no booked total, no approved dates, no approved cabins for a flight or rail segment or no approver leaves the booking unverified. This reads one booking against the authorization behind it: whether a buyer identity on a purchase receipt is the delegating party is wever_delegated-identity-check, a proposed action against a policy text is wever_policy-fit-check and the amount bands an approver covers are wever_approval-chain-check. Verdict travel_authorization_over, travel_authorization_partial, travel_authorization_unverified or travel_authorization_within. POST /api/travel-authorization-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-change-checkAgent Travel: submit the change terms on a booking with the change as it was handled: the terms with whether changes are allowed, the change fee and currency, the hours before the …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,637 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
Description, as the server wrote it Agent Travel: submit the change terms on a booking with the change as it was handled: the terms with whether changes are allowed, the change fee and currency, the hours before the original departure by which a change must be made, whether a name change and a route change are allowed and how a fare difference is settled (pay the difference, pay or refund it, or no difference); optionally the original booking with its reference, traveler, origin, destination, departure and fare; and the change with when it was requested and made, the traveler, origin, destination and departure after it, the new fare, the fee charged and the fare difference charged and refunded. The signed report gives the terms, the original and the change, the hours from the change to the original departure against the window, the fare difference worked out from the two fares with what the rule says to charge or refund beside what was charged and refunded, the fee charged against the fee in the rules, and whether the traveler or route moved against the rules, and counts them, and names a mishandled change: a change the rules leave closed, a change made past the window, a fee above the rules, a fare difference charged other than the two fares give, and a traveler or route changed where the rules keep them; then a lower fare’s difference kept where the rules refund it and a change record that leaves out the new fare, the fee or the difference, with the fix for each and what would change the verdict. No statement of whether changes are allowed, no change fee, no window or no time of the change, no difference rule or no original fare leaves the change unverified. This reads one change to one booking against its fare rules: two versions of a terms text are wever_terms-change-check, a quoted amount against the settled charge is wever_price-truth-check and a purchase’s terms against governing terms are wever_terms-match-check. Verdict travel_change_mishandled, travel_change_partial, travel_change_unverified or travel_change_per_terms. POST /api/travel-change-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-disruption-checkAgent Travel: submit a delayed or cancelled trip with how it was handled: the disruption with its kind (delay, cancellation, schedule change or missed connection), the segment, whe…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,766 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
Description, as the server wrote it Agent Travel: submit a delayed or cancelled trip with how it was handled: the disruption with its kind (delay, cancellation, schedule change or missed connection), the segment, when the carrier knew of it, the original departure and the new departure; the carrier’s stated policy with how soon the traveler is told, how far from the original departure a rebooking may leave, the delay from which the traveler may choose a refund, whether a cancellation gives that choice and the policy as stated; each notice sent with when, the channel and whether it mentions rebooking and a refund; each option offered with its id, kind (rebook, refund, credit or voucher), when it was offered and a rebooking’s departure; and the outcome with the choice, when it was recorded, whether it is recorded against the booking and any refund amount. The signed report gives the disruption with the delay worked out, the policy with whether a rebooking is called for and whether a refund choice is due, the notice with when the first went out against the window and what the notices mention, the options offered with each rebooking’s distance from the original departure and whether the choice was presented, and the outcome, and counts them, and names a mishandled disruption: no notice to the traveler, no rebooking offered where the policy calls for one and no refund-versus-rebook choice where the policy entitles the traveler to it; then notice later than the window, a rebooking outside the window, notices that leave out the options and an outcome missing from the booking’s record, with the fix for each and what would change the verdict. No policy, no notices, no options, no departures to work out the delay or no outcome leaves the handling unverified. This reads one disrupted trip’s handling against the carrier’s stated policy: how a failed agent task is reported is wever_failure-notice-check, the uptime and credit terms in an SLA text are wever_sla-evidence-check and a refund route for a failed paid call is wever_refund-path-check. Verdict travel_disruption_mishandled, travel_disruption_partial, travel_disruption_unverified or travel_disruption_handled. POST /api/travel-disruption-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-fare-checkAgent Travel: submit the fare rules attached to a ticket as purchased with what the buyer was told: the fare rules with the fare class, fare basis, currency, whether the ticket is …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,634 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
Description, as the server wrote it Agent Travel: submit the fare rules attached to a ticket as purchased with what the buyer was told: the fare rules with the fare class, fare basis, currency, whether the ticket is refundable and the refund fee, whether changes are allowed and the change fee, the checked bags and carry-on included, seat selection (included, at extra cost or none), each restriction as filed and the rules as filed; and what the buyer was told, with where (the booking page, a message or a quote), when, the fare class, currency, the same refund, change, baggage and seat terms, the restrictions named and the text, read for wording such as fully refundable, free changes or bags included. The signed report gives the fare rules, what was told, every term point by point with the rules beside the telling, whether they match and whether the buyer was told more or less than the rules give, every restriction with whether the buyer was told of it, and the promising wording found with what the rules contradict, and counts them, and names a misstated fare: a ticket told as refundable that is not, a refund or change fee told lower than the rules, changes told as allowed where the rules allow none, baggage told as included beyond the rules and wording that promises flexibility or bags the rules contradict; then another fare class, seat selection told more favorably, a restriction never told, terms told more strictly than the rules and fees told in another currency, with the fix for each and what would change the verdict. Nothing told, no refundability, no change fee or no baggage allowance in the rules leaves the fare unverified. This reads one ticket’s fare rules against what the buyer was told: a purchase’s price, scope and cancellation against governing terms is wever_terms-match-check, a quote against a vendor’s terms is wever_vendor-terms-match-check and a quoted amount against the settled charge is wever_price-truth-check. Verdict travel_fare_misstated, travel_fare_partial, travel_fare_unverified or travel_fare_as_stated. POST /api/travel-fare-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-itinerary-checkAgent Travel: submit the itinerary as booked with the trip as requested: the request with the traveler, the origin and destination codes, whether a return was requested, the earlie…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,536 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
Description, as the server wrote it Agent Travel: submit the itinerary as booked with the trip as requested: the request with the traveler, the origin and destination codes, whether a return was requested, the earliest departure and latest arrival for the outbound and the return, the shortest workable connection in minutes, the most connections each way and the longest layover accepted in hours (6 when left out); and the booking with its reference, the traveler and each segment with its id, kind, from and to, departure and arrival with their offsets, carrier, number and direction (outbound or return). The signed report gives the request, the booking, the traveler on the request beside the traveler on the booking, each direction with its endpoints, first departure and last arrival read against the requested places and times, and every connection with its place and length in minutes against the minimum, and counts them, and names a mismatched itinerary: another traveler on the booking, a direction that leaves from or arrives at another place, a time outside the request, a segment that leaves from somewhere other than where the one before it arrived, a connection under the shortest workable one and a requested return left unbooked; then a connection within 30 minutes of the minimum, a layover past the longest accepted and more connections than accepted, with the fix for each and what would change the verdict. No traveler on either side, no requested endpoints or times, no minimum connection or connecting segments with no times leave the itinerary unverified. This reads one booked itinerary against the trip requested: a trip planned, booked and held on Wever is read by wever_sports-trip-status and assembled by wever_sports-plan-trip, and a purchase’s terms against its governing terms are wever_terms-match-check. Verdict travel_itinerary_mismatched, travel_itinerary_partial, travel_itinerary_unverified or travel_itinerary_matching. POST /api/travel-itinerary-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_travel-passenger-checkAgent Travel: submit the passenger details on a booking with the traveler’s own records: the traveler with the name on their ID (given, middle and family, or one line), the date of…Can send data out1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.traveler.phone): what the tool writes goes there.
Findings
- Info
The description is 3,632 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
Description, as the server wrote it Agent Travel: submit the passenger details on a booking with the traveler’s own records: the traveler with the name on their ID (given, middle and family, or one line), the date of birth on the ID, the current email and phone, when the contact details were last confirmed, each loyalty number with its program and the known-traveler number; the passenger on the booking with the same name parts, date of birth, email, phone, loyalty numbers and known-traveler number; and optionally the carrier with its name, its loyalty program and whether it needs a date of birth on the booking. The signed report gives each name part on the ID beside the booking with whether it matches and whether the given and family names sit in each other’s fields, the date of birth on each side with whether the carrier needs it, each contact field on record beside the booking, and every loyalty program and the known-traveler number present or absent on each side, and counts them, and names mismatched passenger details: a family or given name spelled differently from the ID, names swapped between fields, a date of birth that differs from the ID and a date of birth missing where the carrier needs one; then a middle name that differs beyond an initial, contact details that differ from the current ones or are missing, a loyalty or known-traveler number missing or differing from the records and a number the records leave out, with the fix for each and what would change the verdict. No ID name, no booked name, no statement of whether a date of birth is needed or no current contact details leave the details unverified. This reads one passenger’s booked details against the traveler’s records: a purchase receipt’s buyer identity against its delegation is wever_delegated-identity-check, a new agent’s identity at joining is wever_onboarding-identity-check and a vendor’s identity is wever_vendor-identity-check. Verdict travel_passenger_mismatched, travel_passenger_partial, travel_passenger_unverified or travel_passenger_matching. POST /api/travel-passenger-check. Reads the travel records supplied in the call and returns a signed report with the verdict, the evidence read, the findings with their fixes, and what would change the verdict; the records stay in the report, with no network request, fetch, message, payment, booking or external write. Agent Travel reads a travel purchase an agent made, after the fact, from the traveler’s own records: the booking against the travel authorization, the itinerary against the trip as requested, the passenger details against the traveler’s ID, the fare rules against what the buyer was told, a change and a cancellation against the fare rules, the extras against what was authorized and needed, and a disruption against the carrier’s stated policy. Beside it, the Sports Trips tools such as wever_sports-trip-status plan, book and read a trip held on Wever, while a travel check reads a purchase made anywhere against its own records; wever_delegated-authority issues and reads the grant itself, while the authorization check reads one booking’s price, dates and class against the travel authorization behind it; and the Tokenized Asset Purchase checks such as wever_tokenized-quote-check read a tokenized asset purchase, while a travel check reads a travel booking. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_treasury-policy-checkAgent Treasury: submit balances (wallet label, asset, chain, amount, and a USD value where you have it) with your stated treasury policy: concentration caps per asset or per chain …No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Treasury: submit balances (wallet label, asset, chain, amount, and a USD value where you have it) with your stated treasury policy: concentration caps per asset or per chain as percentages, a minimum reserve in a named asset, and assets or chains to avoid. The signed report returns a verdict per rule with the computed shares and names every breach. Wallet labels and addresses are read as written and for format only. Verdict compliant, breaches_found or not_computable. POST /api/treasury-policy-check. Reads only the material in the call: no wallet, chain, market, bank or counterparty lookup, and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-treasury-contract. Human page: https://weverlabs.com/agent-treasury/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-aging-checkAgent Triage: submit the queue items with their ages and reviews: each item with its id, its state (open, waiting, in progress or closed), when it entered the queue and when it lef…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,783 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit the queue items with their ages and reviews: each item with its id, its state (open, waiting, in progress or closed), when it entered the queue and when it left, when it was re-reviewed while it waited, and its own review point, such as 2 days; and optionally the review point for any item that states none and the moment to read the queue at. The signed report gives every item with its age, its reviews, its review point, the current wait of an open item and its status as reviewed, reviewed_late, overdue, untouched, unwatched or unconfirmed, counts the items, the open items, the reviews and the items untouched, overdue and reviewed late, and states the review point and the moment read, and names aging ignored: waiting items that have sat untouched past their review point since entering the queue, and waiting items past their review point since their last review; then items that went past the point between reviews and were caught up later, with the fix for each. This reads a queue of items over time rather than the dates inside one document, which is wever_deadline-check: open invoices totaled by overdue bucket are wever_invoice-aging-brief, and open-commitment statuses gone stale on a ledger are wever_commitment-tracking-check. Verdict aging_ignored, aging_partial, aging_unwatched or aging_reviewed. POST /api/triage-aging-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-disposition-checkAgent Triage: submit the queue items with how each one ended, and the items received: each item with its id, its state (open, waiting, in progress or closed), its disposition (acte…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,802 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit the queue items with how each one ended, and the items received: each item with its id, its state (open, waiting, in progress or closed), its disposition (acted on, deferred, declined or merged as a duplicate), the recorded reason for a deferral or a decline, the item a duplicate was merged into, when a deferred item comes back and when it left the queue; and optionally the ids of every item received in the period, from intake. The signed report gives every item with its disposition and its status as acted_on, deferred, declined, duplicate, open, partly_recorded, undisposed or unconfirmed, every item received with its status as recorded or dropped, and counts the dispositions of each kind, the open items and the items dropped, and names items dropped silently: items received with no record, and items that left the queue with no disposition recorded; then deferrals and declines with no reason, duplicates linked to no item in the queue and deferrals with no date to come back, with the fix for each. This reads how every queue item ends: commitments marked done read against their delivery evidence are wever_commitment-closure-check, and where a refused or silent handed-off task goes is wever_handoff-failure-check. Verdict items_silently_dropped, disposition_partial, disposition_unrecorded or dispositions_recorded. POST /api/triage-disposition-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-escalation-checkAgent Triage: submit the escalation thresholds with each queue item’s escalation record: each threshold with its id, the class or severity it covers, how long an item may wait befo…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,133 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit the escalation thresholds with each queue item’s escalation record: each threshold with its id, the class or severity it covers, how long an item may wait before it breaches, such as 4 hours, and the human or higher tier it escalates to; each item with its id, its class, the id of the threshold that covers it, when it entered the queue, when it was handled, whether it was escalated, when and to whom, and whether its context went with it; and optionally how long past the breach an escalation may follow (1 hour when left out) and the moment to read the queue at. The signed report gives every threshold, every item with how long it waited, whether it breached, its escalation and its status as clear, escalated, unescalated, late, no_context, partly_escalated or unconfirmed, and counts the breaches, the escalations, the breaches left unescalated, the late escalations and the escalations with context, and names escalation that comes late or not at all: items that breached their threshold with no escalation and escalations long after the breach; then escalations with no context attached, with no named recipient and to another tier than the threshold names, with the fix for each. This reads a single queue item breaching its threshold and moving up a tier, with its context attached: how a group of agents escalates a disagreement or a stall is wever_coordination-escalation-check, whether an incident escalation policy states triggers an operator could apply is wever_escalation-threshold-check, and where work goes when it hits a limit is wever_breach-path-check. Verdict escalation_late, escalation_partial, escalation_unrecorded or escalation_clean. POST /api/triage-escalation-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-intake-checkAgent Triage: submit the intake records an agent captured for incoming work, with the arrival log behind them: each item with its id, where it came in (an inbox, a form, chat, an A…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,750 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit the intake records an agent captured for incoming work, with the arrival log behind them: each item with its id, where it came in (an inbox, a form, chat, an API or a monitor), when it arrived, who sent it or asked for it, what it is about, when the intake record was complete and when sorting began, the first severity, priority or route set on it; and optionally each arrival as the source systems logged it, with its id, its source and when it arrived. The signed report gives every intake record with the parts it is missing, whether it was sorted before it was complete and its status as complete, incomplete, sorted_early or unconfirmed, every arrival with its status as captured or uncaptured, and counts the items, the arrivals, the arrivals left uncaptured and the items sorted early, and names intake left loose: arrivals with no intake record and items sorted before their intake record was complete; then records with no source, no arrival time, no requester or no subject, with the fix for each. This reads incoming work captured before any sorting begins: promises captured from messages are wever_commitment-capture-check, and the records of handoffs a system made are wever_handoff-record-check. Verdict intake_loose, intake_partial, intake_missing or intake_complete. POST /api/triage-intake-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-priority-checkAgent Triage: submit the queue items with their assessments and the priority set on each: each item with its id, its severity and its impact (critical, high, medium or low: how ser…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,968 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit the queue items with their assessments and the priority set on each: each item with its id, its severity and its impact (critical, high, medium or low: how serious it is and how many people or how much work it reaches), the priority level set on it, 1 handled first, and the recorded reason when the priority departs from the assessments. The signed report gives every item with its assessments, its priority and its status as consistent, overridden, inverted, inconsistent, unassessed or unranked, every pair of items where one that outranks the other on both assessments waits behind it, with whether a reason is recorded, and counts the items ranked and assessed, the inversions and the overrides, and names a priority order that runs against the assessments: an item at least as severe and as wide in impact as another, and more on one of them, that waits behind it with no recorded reason; then items with the same assessments on different priorities, priorities set with no assessment behind them and assessed items left unranked, with the fix for each. This reads the handling order of a triage queue before any promise exists: running agents triaged in priority order from a status snapshot are wever_agent-attention-brief, queue observations turned into an operator brief are wever_agent-daily-operator-brief, and open commitments read against deadlines and capacity are wever_commitment-deadline-check. Verdict priority_arbitrary, priority_partial, priority_unranked or priority_consistent. POST /api/triage-priority-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-report-checkAgent Triage: submit the triage report written for the principal, with the queue records behind it: its title, its audience, the period it covers and its text, read for wording suc…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,084 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit the triage report written for the principal, with the queue records behind it: its title, its audience, the period it covers and its text, read for wording such as nothing falls through the cracks, zero backlog or inbox zero; whether it states the volume received in each class; for the items received, overdue, escalated, disposed and still open, whether the report states them, how many it names and whether each figure links to the queue records behind it; and optionally the record of items received in the period and, of those, overdue, escalated, disposed and still open. The signed report gives the report with any nothing-left-behind wording, each figure with whether it is stated, its count and whether it is backed by the records, and the record with the items unaccounted for, and names a report that inflates the queue record: figures misstated in the direction that flatters the queue (more received or disposed, fewer overdue, escalated or open), figures with no link to the records behind them, and wording such as nothing falls through the cracks while the record holds items unaccounted for, overdue or still open; then figures left out of the report, figures misstated the other way and volume with no breakdown by class, with the fix for each. This reports on an agent’s triage queue against its queue records: a group of agents’ report on its friction is wever_coordination-report-check, a kept-promises report is wever_commitment-report-check and a learning report is wever_learning-report-check. Verdict triage_report_inflated, triage_report_partial, triage_report_missing or triage_report_sound. POST /api/triage-report-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-routing-checkAgent Triage: submit the stated routing rules with where each queue item was routed: each rule with its id, the class of item it routes, the queue and the owner it routes to; and e…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,891 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit the stated routing rules with where each queue item was routed: each rule with its id, the class of item it routes, the queue and the owner it routes to; and each item with its id, the class it was sorted into, the queue and owner it was routed to, the id of the rule applied, the routing reason recorded on the item and how many times it was moved after its first route. The signed report gives every rule, every item with the queue and owner its rule names and its status as routed, overridden, partly_routed, misrouted, unrouted or unexplained, and counts the items routed by rule, misrouted, overridden and carrying a reason, and names misrouted items: items that went to another queue or owner than their rule names with no recorded reason, and items routed to no queue and no owner; then routes with no reason on the item, owners named as a team, someone or TBD, and items moved twice or more after their first route, with the fix for each. An item that departs from its rule with a recorded reason is an override. This reads incoming items assigned to queues and owners, a sorting decision rather than a delegation: why a router or marketplace chose one seller is wever_router-disclosure-check, routing delivered watch fires and job results is wever_alert-route, and the written bounds of one delegation are wever_delegation-scope-check. Verdict misrouted_items, routing_partial, routing_unexplained or routing_sound. POST /api/triage-routing-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_triage-severity-checkAgent Triage: submit a stated severity scale with each queue item’s severity and urgency assessment: each level of the scale (critical, high, medium or low) with the criteria an it…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,939 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
Description, as the server wrote it Agent Triage: submit a stated severity scale with each queue item’s severity and urgency assessment: each level of the scale (critical, high, medium or low) with the criteria an item meets at that level; and each item with its id, what it is about, the severity assessed, the stated criterion the assessment cites, the assessment note as written, read for gut-feel wording such as feels urgent, I think or probably, the urgency assessed on the same scale and its basis, such as a deadline, an outage window or a time limit. The signed report gives the scale, every item with its rating, the level whose criterion it cites, any gut-feel wording and its status as assessed, partly_assessed, guessed, off_scale, mismatched, unassessed or unconfirmed, counts the items rated, assessed, guessed and unassessed, and lists every gut-feel word found, and names severity guessed rather than assessed: ratings at a level the scale does not define, ratings by feel with no stated criterion and ratings that cite another level’s criterion; then items left unrated, ratings with no criterion or one the scale does not state, and urgency missing or stated with no basis, with the fix for each. This reads queue items rated against a stated scale: the severity of failure conditions in alert rules is wever_observability-alert-check, and the severity levels an incident response plan defines are wever_incident-plan-check. Verdict severity_guessed, severity_uneven, severity_unassessed or severity_assessed. POST /api/triage-severity-check. Reads only the recorded queue items and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no item is captured, ranked, routed, escalated or closed and no queue is changed. Agent Triage reads how an agent sorts incoming work, from the item captured at intake to the disposition it ends with: it is not multi-agent coordination (how a group of agents organizes shared work, with its own escalation path, the wever_coordination checks; triage escalation is a single queue item breaching its threshold and moving up a tier, with its context attached), not the promises an agent has made (promises and deadlines once they are made, the wever_commitment checks; triage decides handling order and routing before any promise exists), not delegation (handing work off with authority, the wever_delegation checks; triage assigns incoming items to queues and owners, a sorting decision rather than a delegation) and not the dates inside one document (wever_deadline-check; the triage checks read a queue of items over time). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-abstention-checkAgent Uncertainty: submit an agent’s answers against its own confidence threshold: optionally the policy with the confidence below which the agent declines or defers in percent and…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,765 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit an agent’s answers against its own confidence threshold: optionally the policy with the confidence below which the agent declines or defers in percent and whether that threshold is written where the principal can read it; and each question with its id, the question as asked, the confidence the agent held in percent, what it did (answered, declined or deferred), its reply as given, read for guess wording such as my best guess or off the top of my head, why it declined or deferred as told to the principal, and what would resolve the question as named, such as the signed contract or the Q4 figures. The signed report gives the policy, every question with its status as answered, abstained, answered_below_threshold, guessed, unexplained, unresolved, abstained_above_threshold or unconfirmed, and counts the answers, declines, deferrals and questions below the threshold, and names an agent that guesses where it should abstain: questions answered below its stated threshold, and answers whose own wording calls them a guess; then declines or deferrals that give no reason, declines or deferrals that name nothing that would resolve them, questions declined at or above the threshold and a threshold never written down, with the fix for each. This reads one agent’s discipline in declining below its own threshold: when an operator’s escalation policy fires is wever_escalation-threshold-check, and how a group of agents escalates a disagreement is wever_coordination-escalation-check. Verdict abstention_undisciplined, abstention_gaps, abstention_unverified or abstention_disciplined. POST /api/uncertainty-abstention-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-assumption-checkAgent Uncertainty: submit the assumptions under one result: optionally the result with its statement and its text as delivered, read for assumption wording such as assuming, presum…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,739 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit the assumptions under one result: optionally the result with its statement and its text as delivered, read for assumption wording such as assuming, presumably, given that or all else equal; and each assumption the result lists with its id, the assumption as written, whether the result changes if it fails, whether the result flags it as load-bearing, what changes in the result if it fails and whether it was checked before the result was given. The signed report gives the result with the assumption wording read from it, every assumption with its status as disclosed, unflagged, impact_unstated, unchecked, vague, not_load_bearing or unconfirmed, and counts the load-bearing, flagged, checked assumptions and those with a stated consequence of failure, and names a result that hides what it assumes: assumption wording in the result with no assumptions listed, an assumption the result changes with left unflagged as load-bearing, and a load-bearing assumption that says nothing about what changes if it fails; then a load-bearing assumption never checked and an assumption stated too loosely to check, such as normal conditions or business as usual, with the fix for each. This reads the assumptions under a result an agent gives, not an evaluation’s design or a knowledge base’s sources: the design of a formal evaluation is wever_evaluation-design-check, and the sources behind the facts in a knowledge base are wever_knowledge-source-check. Verdict assumptions_hidden, assumption_disclosure_gaps, assumption_disclosure_unverified or assumptions_disclosed. POST /api/uncertainty-assumption-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-calibration-checkAgent Uncertainty: submit an agent’s stated confidence with its recorded hit rate on comparable past claims: optionally the agent and the kind of claim compared; each confidence ba…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,971 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit an agent’s stated confidence with its recorded hit rate on comparable past claims: optionally the agent and the kind of claim compared; each confidence band with its name, the confidence stated for claims in it in percent, how many past claims in it have a recorded outcome, how many of those proved right and whether those past claims are the same kind as the claims made now; optionally the same comparison period by period, oldest first, with the mean confidence stated, the claims and the claims that proved right; how many past claims a band needs before its hit rate counts (30 when left out); and how many points the hit rate may sit from the stated confidence (10 when left out). The signed report gives every band with its hit rate, its gap in points, its sample as adequate or thin and its status as calibrated, overconfident, underconfident, thin_sample, not_comparable, unrecorded or unconfirmed, and the drift record with the first and latest gaps and how far the gap moved, and names confidence that does not match the record: bands claiming more than the record delivers, bands claiming less, each read as systematic when two or more bands run the same way, and a gap that moved past the tolerance from the first period to the latest; then bands resting on too few past claims and bands compared against claims of another kind, with the fix for each. This compares an agent’s own stated confidence with its own record: whether an evaluation’s claimed figures match its measured results is wever_evaluation-result-check, drift between two evaluation runs of one agent is wever_evaluation-drift-check, and runtime success rates and latencies are wever_observability-metric-check. Verdict confidence_miscalibrated, calibration_gaps, calibration_unverified or confidence_calibrated. POST /api/uncertainty-calibration-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-evidence-gap-checkAgent Uncertainty: submit the claims in one work product with the evidence they cite: each claim with its id, the claim as written, the confidence stated for it in percent, whether…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,941 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit the claims in one work product with the evidence they cite: each claim with its id, the claim as written, the confidence stated for it in percent, whether the answer or decision rests on it and the ids of the evidence items it cites; each evidence item with its id, where it comes from and the support it gives as recorded (strong, moderate, thin or none) and whether it bears on the claim directly rather than by analogy; and optionally the highest confidence strong, moderate and thin support carry (95, 80 and 60 when left out, with no supporting evidence carrying at most 50). The signed report maps every claim to its cited evidence with its best support, indirect evidence counted one level lower, the ceiling that support carries and the margin between the stated confidence and that ceiling, lists every evidence item with the claims that cite it, and gives each claim as backed, exceeds_evidence, load_bearing_unsupported, load_bearing_thin, uncited, indirect_only or unconfirmed, and names confidence that outruns the evidence: claims stating more confidence than their cited evidence supports, load-bearing claims resting on no evidence, and load-bearing claims resting on thin evidence only; then other claims citing no evidence and claims resting only on indirect evidence, with the fix for each. The support each item gives is taken as recorded and the sources are not graded: marking each part of a claim supported or contradicted by evidence text is wever_claim-evidence-check, verifying a data sample field by field is wever_data-evidence-check, and whether a cited passage makes a claim is wever_knowledge-grounding-check. Verdict confidence_outruns_evidence, evidence_backing_gaps, evidence_backing_unverified or confidence_backed_by_evidence. POST /api/uncertainty-evidence-gap-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-freshness-checkAgent Uncertainty: submit how one answer discloses the age of what it knows: optionally the answer with its title, when it was given, the knowledge cutoff it rests on and whether i…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,946 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit how one answer discloses the age of what it knows: optionally the answer with its title, when it was given, the knowledge cutoff it rests on and whether it states it, the date it speaks as of and whether it states it, and its text, read for present-tense wording such as currently or the latest; each fact it states with its id, the fact as stated, whether it is time-sensitive (a price, a rate, an officeholder or a version), when it was last checked against its source, whether it was checked again for this answer, whether the answer flags it as possibly out of date and how many days it stays current (30 when left out); and each source it cites with its id, the source, when it was published or updated, how many days it stays useful (365 when left out) and whether the answer marks it as past its useful life. The signed report gives the answer with its dates and present-tense wording, every fact with its age from its last check or else from the cutoff and its status as current, reverified, flagged_stale, stale_unflagged, not_time_sensitive or unconfirmed, and every source with its age and status, and names an answer that hides how old its knowledge is: a knowledge cutoff not stated, time-sensitive facts past their shelf life neither re-verified nor flagged, and sources past their useful life cited without a mark; then an as-of date left out, including beside present-tense wording, with the fix for each. This reads one answer’s disclosure of its own staleness: whether a knowledge base serves its entries as current is wever_knowledge-freshness-check, and whether feedback relied on now is still current is wever_feedback-freshness-check. Verdict staleness_undisclosed, staleness_disclosure_gaps, staleness_disclosure_unverified or staleness_disclosed. POST /api/uncertainty-freshness-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-range-checkAgent Uncertainty: submit the estimates in one work product: each estimate with its id, what is estimated, the single figure given, the low and high bounds of its range, the unit, …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,601 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit the estimates in one work product: each estimate with its id, what is estimated, the single figure given, the low and high bounds of its range, the unit, the figure as written (such as 1,234,567 units or 1.2 million), why the bounds sit where they do as stated, and the evidence behind it (strong, moderate or thin). The signed report gives every estimate with its range, its width as a share of its midpoint, the precision its written figure claims and its status as sound, point_only, too_narrow, bounds_unreasoned, overstated_precision, too_wide or unconfirmed, counts the ranges, single figures and reasoned bounds, and gives the width rules, and names estimates that claim more precision than they have: a single figure with no range, and a range narrower than its evidence supports (at least 50% of its midpoint on thin evidence and 20% on moderate evidence); then bounds with no stated reasoning, a figure written finer than one hundredth of its range, and a range wider than 100% of its midpoint on strong evidence, with the fix for each. This reads the honesty of the ranges an agent gives, not whether the figures are right: whether a benchmark claim’s figures appear in its cited source is wever_benchmark-evidence-check, and the spread around an evaluation’s measured results is wever_evaluation-result-check. Verdict estimates_falsely_precise, estimate_range_gaps, estimate_range_unverified or estimate_ranges_sound. POST /api/uncertainty-range-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-report-checkAgent Uncertainty: submit the uncertainty report written for the principal, with the record behind it: its title, its audience, the answer as the report states it and optionally it…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 3,007 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit the uncertainty report written for the principal, with the record behind it: its title, its audience, the answer as the report states it and optionally its text, read for certainty wording such as beyond doubt or definitely; whether it states what is known, what is unknown, what is assumed, what is contested and the evidence that would change the answer, where each sits (beside the answer, in a later section, in an appendix or in a separate document) and how many of each it names; whether the uncertainty sits apart from the answer rather than woven through it; and optionally the record of established points, open unknowns, assumptions, contested points and evidence that would change the answer. The signed report gives the report with any certainty wording, the disclosure of each kind of uncertainty with its placement and count and whether it sits apart from the answer, and the record, and names a report that conceals the uncertainty behind the answer: unknowns left out or stated as none while the record holds some, assumptions left out, contested points left out, and uncertainty woven through the answer so the principal cannot tell the answer from what qualifies it; then what is known or what would change the answer left out, counts below the record’s, unknowns, assumptions or contested points in an appendix or a separate document, and certainty wording while the record holds unknowns or contested points, with the fix for each. What a knowledge base holds and lacks across topics is wever_knowledge-report-check, an evaluation summary of scores is wever_evaluation-report-check, an observability summary is wever_observability-report-check, and a feedback summary is wever_feedback-report-check. Verdict uncertainty_concealed, uncertainty_report_gaps, uncertainty_report_unverified or uncertainty_report_sound. POST /api/uncertainty-report-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_uncertainty-statement-checkAgent Uncertainty: submit the confidence statements in one work product, claim by claim: optionally the work with its title and its text as delivered, read for one blanket confiden…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 2,843 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
Description, as the server wrote it Agent Uncertainty: submit the confidence statements in one work product, claim by claim: optionally the work with its title and its text as delivered, read for one blanket confidence for the whole of it such as high confidence overall or I am 90% confident in this answer; optionally that blanket confidence as written; and each claim with its id, the claim as written, whether it is uncertain (an estimate, a forecast or an inference rather than a checked fact), the confidence stated for it as written (such as about 70%, 3 in 4 or likely) and whether that confidence sits beside the claim it modifies. The signed report gives the work with its blanket confidence and whether it was stated or read from the wording, every claim with its stated confidence, the figure read from it and its status as stated, missing, blanket_only, unattached, unquantified, worded_certain, settled or unconfirmed, and counts the uncertain, quantified and attached claims, and names a work that leaves its uncertainty unstated: uncertain claims with no confidence of their own, uncertain claims covered only by one blanket confidence for the whole answer, and confidence placed away from the claim it modifies; then confidence in words only with no figure or range, and certainty wording such as definitely on a claim the work marks as uncertain, with the fix for each. This reads what an agent says about its own confidence, not how good the output is: scoring an output against a rubric is wever_evaluation-scoring-check, and whether a cited passage makes a claim is wever_knowledge-grounding-check. Verdict confidence_unstated, confidence_statement_gaps, confidence_statement_unverified or confidence_stated. POST /api/uncertainty-statement-check. Reads only the recorded work product and records in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no answer is changed, no source is fetched and no party is contacted. Agent Uncertainty reads the uncertainty an agent carries in its own work products and answers, its stated confidence, what it does not know and what it assumed: it is not formal scoring of outputs against rubrics, datasets and benchmarks (the wever_evaluation checks), not grading the source material itself (wever_claim-evidence-check, wever_data-evidence-check and the other evidence checks), not traces, metrics and runtime behavior (the wever_observability checks) and not the loop around feedback received (the wever_feedback checks). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_unified-agent-checkoutPaid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days.Reads text others write1 finding
Why it is in this group
- The name "wever_unified-agent-checkout" says "checkout": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Findings
- Info
The description mentions "credentials".
What this check looks forMatched text
description: "...nified-agent-checkout. Existing product credentials, signed mandates, and single-use action..."
Description, as the server wrote it Paid URL: 0.10 USDC per call on Base; 10 signed-wallet free calls shared over rolling 30 days. Read the backend 402 body for x402 terms. Itemize caller-supplied USD estimates and calculate a total for supported computations. Operating boundary: Supply computation items and your USD estimates. Read the itemized total. An agent carries the estimate into its work plan; a human compares the items before requesting a provider checkout. POST /api/unified-agent-checkout. Existing product credentials, signed mandates, and single-use action grants remain required where applicable. This adapter grants no authority and never supplies server credentials. This current operation is a bounded computation on caller-supplied data.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_vendor-delivery-checkAgent Vendors: submit the delivery promise for one vendor engagement and where else it is stated: the promise as written, read for hedges such as usually, as soon as possible or es…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,114 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
Description, as the server wrote it Agent Vendors: submit the delivery promise for one vendor engagement and where else it is stated: the promise as written, read for hedges such as usually, as soon as possible or estimated, its timeframe, the date it is delivered by, what the timeframe counts from, the delivery method, what counts as delivered and the proof of delivery the buyer receives; and each other statement of the same promise with its id, its source (the quote, the terms, a listing, the order, an invoice, correspondence or other), its timeframe, delivery date, method and text. The signed report gives the promise read in days (business days at five to the week, a range at its upper end) with any hedges, every other statement with its days, date and method and its status as agrees, contradicts or unread, each part with its status, and counts the statements agreeing, contradicting and unread, and names a delivery promise contradicted where it is stated elsewhere: another number of days, another date or another method; then no timeframe, a timeframe with no number or date, a hedged timeframe, no starting point, no method, nothing saying what counts as delivered and no proof of delivery, with the fix for each. This reads the whole delivery promise before the spend: a shipping promise against its delivery evidence after the fact is wever_shipping-check, delivered work against its acceptance criteria is wever_deliverable-check and an uptime commitment after an outage is wever_sla-credit-check. Verdict delivery_contradicted, delivery_partial, delivery_unstated or delivery_terms_clear. POST /api/vendor-delivery-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_vendor-identity-checkAgent Vendors: submit a vendor as it presents itself and the identity evidence in hand: the vendor’s name, legal name, trading name, web domain, contact email, phone, postal addres…Can send data out1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- outboundThe input takes a destination (input.vendor.phone): what the tool writes goes there.
Findings
- Info
The description is 4,316 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
Description, as the server wrote it Agent Vendors: submit a vendor as it presents itself and the identity evidence in hand: the vendor’s name, legal name, trading name, web domain, contact email, phone, postal address and registration number; and each piece of evidence with its id, its kind (a registration, a contract, an invoice, the website, the payment account, correspondence, a directory or marketplace listing, or other), where it came from, the seller name, domain, email and registration number it shows, the account holder a payment would go to and when it was read. The signed report gives the vendor with its name read without legal suffixes, its domain and its contact domain, every piece of evidence with what it shows and its status as matches, mismatch, listing_only or unread, and counts the evidence matching, mismatched, listed apart and unread, and names a vendor that is not who it claims to be: evidence naming another seller, showing another domain, paying another account holder or carrying another registration number, and a contact address on another domain; then a contact on a free-mail domain, no contact path, no domain, no legal name and no registration entry or signed contract, with the fix for each. Directory and marketplace listings are listed apart and never counted as identity evidence; a vendor with only listings behind it is unverified. This reads one vendor before a spend: an agent’s own identity record is wever_agent-identity-check, a research brief across directory, index and evidence sources is wever_counterparty-brief, an onboarding packet’s completeness and formats are wever_counterparty-onboarding-check and the buyer named on a receipt is wever_delegated-identity-check. Verdict vendor_misrepresented, identity_partial, vendor_unverified or vendor_verified. POST /api/vendor-identity-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_vendor-lockin-checkAgent Vendors: submit the lock-in and exit terms for one vendor engagement and the agent’s summary to its principal: the terms as written, read for wording that renews on its own, …1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,325 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
Description, as the server wrote it Agent Vendors: submit the lock-in and exit terms for one vendor engagement and the agent’s summary to its principal: the terms as written, read for wording that renews on its own, the minimum term, whether it renews automatically, the notice to stop a renewal, how to cancel, the early termination fee, data or asset export with whether it is available, its formats and any fee, formats only this vendor reads and the costs of switching with their amounts; and the summary’s text, read for wording such as cancel anytime or no lock-in, with the lock-in factors it names plainly. The signed report gives each of eight factors (automatic renewal, minimum term, renewal notice, cancellation path, termination fee, data export, proprietary formats, switching costs) as present, absent or unstated, whether it is high and whether the summary names it, and counts the factors read, present, high and named, and names high lock-in left unstated: a term of 12 months or more, 30 days or more of renewal notice, cancellation only by written notice, post or phone, an early termination fee, no export, a paid one or one only in formats this vendor reads, proprietary formats or switching costs that the summary leaves out, and a summary calling the exit free while the terms tie the buyer in; then each high factor named, lower lock-in the summary leaves out and exit facts the terms leave unsaid, with the fix for each. This reads the exit picture before the agent commits: the renewal notice deadline arithmetic in an agreement is wever_renewal-notice-check, export claims against the pages offered for them are wever_export-evidence-check and recurring charges already running are wever_subscription-audit. Verdict lockin_high_unstated, lockin_partial, lockin_unexamined or lockin_low_and_named. POST /api/vendor-lockin-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_vendor-price-consistency-checkAgent Vendors: submit every place the agent can see this vendor’s price for one item: the item, and each source with its id, its kind (a quote, a listing, an invoice, a prior charg…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,322 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
Description, as the server wrote it Agent Vendors: submit every place the agent can see this vendor’s price for one item: the item, and each source with its id, its kind (a quote, a listing, an invoice, a prior charge, an order, a contract or other), the amount it shows, its currency, the quantity the amount is for, when it was issued or charged, its reference and the vendor’s reason for any difference; and optionally the source to compare against (the first quote when left out), a tolerance as a percent of the price (0 when left out) and the agent’s own note on the price, read for wording that averages a difference away such as about, roughly or on average. The signed report gives the source compared against, every source with its price per unit, its difference in amount and percent and its status as compared_against, consistent, explained, unexplained, other_currency or unpriced, and counts the sources consistent, explained, unexplained, in another currency and unpriced with the lowest and highest price per unit, and names a price that is not consistent: differences with no reason from the vendor and a note that averages the differences away rather than naming each; then differences explained by the vendor, prices in another currency with no rate and sources with no price, with the fix for each. Every unexplained difference is named rather than averaged away. This reads the price this vendor quoted this buyer before the spend: one paid call’s quote against its settled charge is wever_price-truth-check, an invoice against its job or mandate is wever_invoice-check, a pricing page priced two ways is wever_pricing-evidence-check and recurring charges already running are wever_subscription-audit. Verdict price_inconsistent, price_variance_noted, price_uncompared or price_consistent. POST /api/vendor-price-consistency-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_vendor-quote-checkAgent Vendors: submit the quote in front of the agent: its id, the vendor, the item or service, the scope, the quantity, the unit price, the currency, the total, when it was issued…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,227 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
Description, as the server wrote it Agent Vendors: submit the quote in front of the agent: its id, the vendor, the item or service, the scope, the quantity, the unit price, the currency, the total, when it was issued and when it lapses, its text, read for wording that implies a fee such as plus applicable fees, taxes extra or shipping calculated at checkout and wording that leaves the price open such as starting at, subject to change or estimated, and each priced line with its id, label, kind (an item, a fee, tax, shipping, a discount, a deposit or other) and amount; and optionally the moment to read it at. The signed report gives the quote, every line, each part (item or service, quantity or scope, total price, currency, validity window, fees named) with its status, and counts the lines, the fee lines and the implied fees with the lines’ total beside the stated total, and names a quote too vague to accept: no item or scope, no price, fees implied by wording no line prices, a price left open, lines that do not add up to the total and a total that differs from the unit price times the quantity with nothing itemizing the difference; then no stated total, no currency, no quantity, no validity date, an expired quote and an unpriced line, with the fix for each. Amounts compare to the cent. This reads one quote before it is accepted: ranking two or more quotes for the same job is wever_quote-compare, comparing vendor proposals across price, term and service levels is wever_procurement-decision-brief, a proposed deal’s terms deliverable by deliverable are wever_negotiation-terms-check and conversion quotes are wever_exchange-quote-check. Verdict quote_vague, quote_partial, quote_missing or quote_complete. POST /api/vendor-quote-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_vendor-record-checkAgent Vendors: submit the vendor report sent to the principal, with the vendor record behind it: its title, the vendor it reports on, the engagement it covers and its text, read fo…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,207 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
Description, as the server wrote it Agent Vendors: submit the vendor report sent to the principal, with the vendor record behind it: its title, the vendor it reports on, the engagement it covers and its text, read for wording such as fully vetted, no risks, no hidden fees or best price; for the vendor’s identity standing (verified, partial, unverified or misrepresented), the total cost, the contradictions between the terms and the quote and the risks found, whether the report states each, what it states and whether each links to the vendor record behind it; and optionally the vendor record’s identity standing, total cost, contradictions and risks. The signed report gives the report with any wording that calls the vendor safe, each part with whether it is stated, what it states and whether it is backed by the record, and the record, and names a report that flatters the vendor record: a better identity standing, a lower cost or fewer contradictions or risks than the record shows, parts with no link to the record, and wording such as fully vetted or no risks while the record shows the vendor unverified, contradictions or risks; then parts left out, parts misstated the other way and a report that does not name its vendor, with the fix for each. Costs compare to the cent. This reports one vendor to its principal against the vendor record behind each claim, rather than asserted: normalizing several vendor proposals into a recommendation is wever_procurement-decision-brief, a research brief on a counterparty is wever_counterparty-brief and a report on a goal is wever_goal-report-check. Verdict vendor_report_inflated, vendor_report_partial, vendor_report_missing or vendor_report_sound. POST /api/vendor-record-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_vendor-terms-match-checkAgent Vendors: submit the quote or order in front of the agent and the vendor’s terms: the quote with its id, price, currency, days to delivery, days to ask for a refund, renewal (…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Info
The description is 4,381 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
Description, as the server wrote it Agent Vendors: submit the quote or order in front of the agent and the vendor’s terms: the quote with its id, price, currency, days to delivery, days to ask for a refund, renewal (none, manual or automatic) and commitment in months; and the terms with where they came from, their version, when they were read, the same six points as the terms set them and their text, read for wording that overrides a quote such as prices subject to change, these terms prevail or quotes are not binding, wording that refuses refunds such as all sales final and wording that renews on its own. The signed report gives the quote, the terms with any overriding clause, each point with what was quoted, what the terms set and its status as match, contradicts, terms_better, terms_silent, not_quoted or unread, and counts the points compared, matching, contradicted, better for the buyer and left unstated, and names terms that contradict the quote: a higher price, another currency, slower delivery, a shorter refund window, a more automatic renewal or a longer commitment than quoted, a clause that lets the terms override the quote, a clause refusing the refund the quote offers and a clause renewing what was quoted as one-off; then quoted points the terms do not state, a renewal or commitment the quote never mentioned and terms with no source, with the fix for each. Each contradiction is named where it costs money. This reads one vendor’s terms against the quote before the spend: purchases already made checked against their governing terms are wever_terms-match-check, two versions of the same terms are wever_terms-change-check, which clauses a terms page holds is wever_terms-evidence-check and who pays when an agent errs is wever_liability-terms-check. Verdict terms_contradict_quote, terms_match_partial, terms_unexamined or terms_match_quote. POST /api/vendor-terms-match-check. Reads only the recorded vendor material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write; no vendor is contacted, no quote is accepted, no order is placed and no spend is approved. Agent Vendors reads the seller on the other side of a spend before the agent commits: whether the vendor is who it claims to be, whether its quote and terms hold together, whether its price stays the same and what it costs to leave. It is not one service’s terms on their own (wever_service-terms-check reads one service’s terms by themselves; the vendor terms-match check reads a vendor’s terms against the specific quote or order in front of the agent, where a contradiction costs money), not pricing structure or fee evidence (wever_pricing-tier-check and wever_fee-evidence-check; the vendor price check reads whether the price this vendor quoted this buyer stays consistent across the quote, the listing and any invoice or prior charge), not single promise types (wever_warranty-check and wever_shipping-check; the vendor delivery and refund checks read the whole delivery promise and the whole remedy path for one vendor engagement), not listing accuracy (wever_catalog-listing-check reads whether a catalog listing is accurate; a directory or marketplace listing being present is not vendor verification, and no vendor check treats listing presence as identity evidence), not after the payment (wever_purchase-record-check and wever_double-charge-check run after a payment; the vendor checks run before the spend, while walking away is still free), not the buyer’s own permission (wever_spending-policy-check and wever_authority-scope-check read the buyer’s permission to spend; the vendor checks read the seller on the other side of the spend: permission is not diligence, and diligence is not permission), not a promise once made (the wever_commitment checks; a vendor engagement becomes a commitment only after the agent accepts it, and the vendor checks guard the moment before that) and not an objective (the wever_goal checks read the objective the work serves; a vendor check reads a counterparty, not an objective). Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_volume-quality-checkAgent Accountability: submit a payments activity report or metrics definition: each headline number with what it measures (transaction count, settled value, payer count, average ti…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Agent Accountability: submit a payments activity report or metrics definition: each headline number with what it measures (transaction count, settled value, payer count, average ticket), its value, how it is computed and what it excludes. The signed report says whether settled value is separated from transaction churn, whether self-payments and test payments are excluded, whether payers are counted as distinct wallets paying with their own money, and every headline number that can rise with no commerce behind it. A metrics summary’s ticket sizes are wever_settled-value-check. Verdict inflatable, quality_unverified or commerce_backed. POST /api/volume-quality-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-accountability-contract. Human page: https://weverlabs.com/agent-accountability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_wallet-activity-briefWever Briefs: plain-language summary of a Base address’s recent USDC flows from public on-chain Transfer logs.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Wever Briefs: plain-language summary of a Base address’s recent USDC flows from public on-chain Transfer logs. Inputs: address and optional window_days (default 30, maximum 90). Returns totals in and out, net flow, the largest transfers, the counterparty address count with the top counterparties by volume, and the exact time window covered. Counterparties appear as on-chain addresses. POST /api/wallet-activity-brief. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns one brief with an Ed25519 signed receipt. Contract: /api/briefs-contract. Human page: https://weverlabs.com/briefs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_wallet-statementWever Briefs: a monthly USDC statement for one Base address, the expense close pack.Reads text others writeNo findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Wever Briefs: a monthly USDC statement for one Base address, the expense close pack. Input address (the form wever_wallet-activity-brief accepts) and month (YYYY-MM, the current month to date or one of the twelve before it). The signed statement gives total USDC in and out on Base, the net change, transfer counts, up to 10 largest transfers with dates and counterparty addresses, and the opening and closing balances where the readers establish them, laid out as plain statement lines for a person or an accountant. It reads the public Base USDC Transfer logs and balance through the wallet activity brief readers. POST /api/wallet-statement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-records-contract. Human page: https://weverlabs.com/briefs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_warranty-checkDiligenceOps: submit a warranty claim (the issue, purchase date and optionally the delivery date, claim date, part, cause, remedy and a stated warranty length) and the warranty ter…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it DiligenceOps: submit a warranty claim (the issue, purchase date and optionally the delivery date, claim date, part, cause, remedy and a stated warranty length) and the warranty terms. The signed report counts the applicable period from purchase or delivery to an end date, says whether the claim falls inside it, whether the coverage names the issue or covers defects in general, which exclusions the cause or issue meets, and whether the remedy asked for is offered, listing each period, coverage and exclusion mismatch with quotes. POST /api/warranty-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-terms-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_watch-createWever Watch: register a standing instruction for the paying wallet.Reads text others writeCan send data outNo findings
Why it is in this group
- The name "wever_watch-create" says "create": a word for a change.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe input takes a URL (input.callback_url): a request to a URL can carry data to whoever runs it.
Description, as the server wrote it Wever Watch: register a standing instruction for the paying wallet. Give a target (a public HTTPS JSON data endpoint, a prediction market as kalshi:<ticker> or polymarket:<id or slug>, or a Base wallet address with USDC or ETH), a condition (field, operator above, below, at_or_above or at_or_below, and a threshold), a callback_url and an expires_at up to 30 days ahead. The server checks on its scheduler every check_interval_seconds (default 300). When the condition is met it POSTs a signed fire report to callback_url and stores the fire record. It re-arms after the condition clears and completes after max_fires (default 1). Version 1 sends reports and moves no money. POST /api/watch-create. One call covers every check until the watch ends. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/watch-contract. Human page: https://weverlabs.com/watch/.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
Only reads 23Show 23 read-only tools
boutique-couponGet the free, shareable code for 10 percent off merchandise at all five Rebecca Belle Boutique stores through Stripe checkout.No findings
Why it is in this group
- The description's first verb is "get": a word for reading.
Description, as the server wrote it Get the free, shareable code for 10 percent off merchandise at all five Rebecca Belle Boutique stores through Stripe checkout. Returns the code, plain terms, redemption steps and store links. Agents may share this code freely with any human. This tool is free and needs no account, wallet or x402 payment.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_atlas-asset-searchSearch covered tokenized assets by literal text, category, stage or six rights filters and return paginated summaries with stage and rights.Reads text others writeNo findings
Why it is in this group
- The name "wever_atlas-asset-search" says "search": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Search covered tokenized assets by literal text, category, stage or six rights filters and return paginated summaries with stage and rights. Empty arguments list all covered assets within the result limit. Reads the versioned Atlas reference dataset, not live market quotes. Each result preserves source dates and labels for provisional or unverified historical display data. POST /api/atlas-asset-search. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof must bind to the exact JSON arguments and this API path, not /mcp. The adapter forwards only caller-supplied proof and payment headers. Authority is read-only: no dataset or external writes, asset trading, autonomous source fetches, spending authority or publication. API access billing is separate: it records free usage or settles an explicit caller-supplied payment authorization. Read /api/atlas-contract and the backend 402 body for exact schemas and billing terms.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_atlas-compareCompare two Atlas assets by id or ticker.Reads text others writeNo findings
Why it is in this group
- The name "wever_atlas-compare" says "compare": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Compare two Atlas assets by id or ticker. Returns seven side-by-side rows: taxonomy stage, issuer involvement, custody and backing, distributions, voting, legal status and redemption. Each row includes the value, detail and source identifiers for both assets. Each new execution reads https://atlas.weverlabs.com/api/assets and returns the dataset version, catalog counts and retrieval evidence. POST /api/atlas-compare. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. Read-only rights analysis and API access billing are separate: the billing layer records shared free usage or settles the caller-supplied payment authorization. Read /api/atlas-rights-contract and the backend 402 body for schemas and billing terms.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_atlas-rights-gapList Atlas rights fields marked not disclosed, with the field detail, source identifiers and sources checked evidence.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "list": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it List Atlas rights fields marked not disclosed, with the field detail, source identifiers and sources checked evidence. Filter by asset or field and paginate the current catalog. Each new execution reads https://atlas.weverlabs.com/api/assets and returns the dataset version, catalog counts and retrieval evidence. POST /api/atlas-rights-gap. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. Read-only rights analysis and API access billing are separate: the billing layer records shared free usage or settles the caller-supplied payment authorization. Read /api/atlas-rights-contract and the backend 402 body for schemas and billing terms.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_boutique-list-productsBrowse Rebecca Belle Boutique products with stable IDs, names, prices, photos, availability and descriptive details.Reads text others writeNo findings
Why it is in this group
- The name "wever_boutique-list-products" says "list": a word for reading.
What it can reach
- untrusted-inputThe description's first sentence says "browse": it reads web pages, which anyone can write.
Description, as the server wrote it Browse Rebecca Belle Boutique products with stable IDs, names, prices, photos, availability and descriptive details. Supply store=women, men, kids, home, or accessories to browse a live collection with query, size, brand, category, min_price, max_price, nwt, sort, page, and page_size. Home offers home goods. Accessories offers shoes, jewelry and accessories. Product IDs use rb-#####, mm-#####, lc-#####, rh-#####, and rs-#####; cross-listed accessories retain their original product IDs and shared stock. The collections share a cart with $8 shipping per whole order and card payment through the boutique checkout. Collection and product records expose agent_purchase_enabled. Omit store and filters to read the legacy nine-piece seasonal snapshot. Catalog prices use 1 USD = 1 USDC; checkout calculates shipping and tax. Anonymous catalog reads are free.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_boutique-product-detailRead one Rebecca Belle Boutique product by its exact catalog ID.No findings
Why it is in this group
- The description's first verb is "read": a word for reading.
Description, as the server wrote it Read one Rebecca Belle Boutique product by its exact catalog ID. Supply store=women, men, kids, home, or accessories to browse a live collection with query, size, brand, category, min_price, max_price, nwt, sort, page, and page_size. Home offers home goods. Accessories offers shoes, jewelry and accessories. Product IDs use rb-#####, mm-#####, lc-#####, rh-#####, and rs-#####; cross-listed accessories retain their original product IDs and shared stock. The collections share a cart with $8 shipping per whole order and card payment through the boutique checkout. Collection and product records expose agent_purchase_enabled. Omit store and filters to read the legacy nine-piece seasonal snapshot. Catalog prices use 1 USD = 1 USDC; checkout calculates shipping and tax. Anonymous catalog reads are free.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_contract-diff-checkDiligenceOps: submit two versions of a terms or contract text (before and after, plain or HTML, up to 256 KB each) and get a signed report of the clauses added, removed or material…Reads text others writeNo findings
Why it is in this group
- The name "wever_contract-diff-check" says "diff": a word for reading.
What it can reach
- untrusted-inputThe description's first sentence says "html": it reads web pages, which anyone can write.
Description, as the server wrote it DiligenceOps: submit two versions of a terms or contract text (before and after, plain or HTML, up to 256 KB each) and get a signed report of the clauses added, removed or materially changed, grouped by topic (fees, refund, cancellation, ownership, liability, term), each quoted briefly with its offset. A change is material when a figure, an obligation word, a negation or a topic changes. Wording-only edits are counted separately. POST /api/contract-diff-check. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_digest-statusStanding jobs: one digest owned by the calling wallet: schedule, windows closed, posted and skipped as empty, items waiting in the open window, recent windows with delivery outcome…No findings
Why it is in this group
- The name "wever_digest-status" says "status": a word for reading.
Description, as the server wrote it Standing jobs: one digest owned by the calling wallet: schedule, windows closed, posted and skipped as empty, items waiting in the open window, recent windows with delivery outcomes, and one full signed digest report (the latest, or window_index). POST /api/digest-status. Standing jobs check and report: they run read-and-report Wever tools only and never move money, place orders or change state. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/standing-jobs-contract. Human page: https://weverlabs.com/standing-jobs/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_directory-statusAgent directory: the state of a directory listing or submission.Reads private dataReads text others writeCan send data outNo findings
Why it is in this group
- The name "wever_directory-status" says "status": a word for reading.
What it can reach
- private-dataThe name "wever_directory-status" says "status" and "directory": it reads files.
- private-dataThe description's first sentence says "directory": it reads files.
- untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Claude Code: ask, because it is tagged outbound.
Description, as the server wrote it Agent directory: the state of a directory listing or submission. Input submission_id, or url as a listed website or MCP endpoint. The signed report states live (with its layer, its review state, any completed checks, the auth label when its MCP handshake passed anonymously and an earned verification date), queued, in operator review (with the checks to fix) or not found, read only from the directory listings and its review queue. POST /api/directory-status. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-records-contract. Human page: https://weverlabs.com/directory/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wever_dispute-statusRead the dispute and full operator-signed receipt chain, including opening, each evidence submission, extension and resolution or timeout.Reads text others writeNo findings
Why it is in this group
- The name "wever_dispute-status" says "status": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Read the dispute and full operator-signed receipt chain, including opening, each evidence submission, extension and resolution or timeout. Read-only: no settlement, signing or reconciliation. Requires a party wallet proof. POST /api/dispute-status. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path. Fees are separate from the locked principal and grant no resolution authority. Disputes expire after 14 days; unresolved expiry refunds the payer with a signed timeout receipt, subject to settlement confirmation and the UTC daily cap. External custody and security audit remains pending. Company-authored. Read /api/dispute-contract for schemas, operator authorization and recovery limits.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_mandate-statusRead live, halted, revoked, expired or replaced status for an exact version.Reads text others writeNo findings
Why it is in this group
- The name "wever_mandate-status" says "status": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Read live, halted, revoked, expired or replaced status for an exact version. Free. POST /api/mandate-status. Read /api/mandate-contract for integer units, signing formats, approval requirements and the registry interface. Issue and amend use the existing x402 payment rail. Status, halt and revoke have no payment requirement. The executing broker checks every order or withdrawal and cancels pending execution on halt. Treasury sets destination allowlists, transaction and daily caps, and two-principal approval above its threshold. The registrar timestamps versions and chain fingerprints. Human page: https://weverlabs.com/trading-mandate/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_observability-log-checkAgent Observability: submit an agent’s logging design, a sample of its log, or both: the design as whether each record names the actor, the principal and the authorization behind t…Reads private dataReads text others writeNo findings
Why it is in this group
- The name "wever_observability-log-check" says "log": a word for reading.
What it can reach
- private-dataThe description's first sentence says "message": it reads messages.
- untrusted-inputThe description's first sentence says "message": it reads messages, which other people can write.
Description, as the server wrote it Agent Observability: submit an agent’s logging design, a sample of its log, or both: the design as whether each record names the actor, the principal and the authorization behind the action, holds the step’s inputs and outputs and carries a timestamp; each sample entry with its id, time, action, kind (a read, a write, a payment, a message, a tool call, a decision or other), actor, principal, authorization and its recorded inputs and outputs; and optionally the grants, mandates and approvals on file. The signed report says field by field what the design promises and what the entries hold, and names what an audit would need and not find: which actor took each action, the authorization behind it, and the inputs and the outputs of each step; then records with no principal or timestamp, shared actors such as system or agent that name no one actor, authorizations not on file and a design the sample log does not follow, with the fix for each. Whether each step of one run names its actor is wever_actor-attribution-check, whether an approval record holds up is wever_approval-record-check and a counterparty’s SOC 2 or ISO audit report is wever_audit-evidence-check. Verdict audit_blind, log_gaps, log_unverified or audit_ready. POST /api/observability-log-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-observability-contract. Human page: https://weverlabs.com/agent-observability/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_payout-statusRead a payout and its signed transition receipts by payout_id or the original create idempotency_key.Reads text others writeNo findings
Why it is in this group
- The name "wever_payout-status" says "status": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Read a payout and its signed transition receipts by payout_id or the original create idempotency_key. Read-only: no payout state change, signing, reconciliation or transaction submission. POST /api/payout-status. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. Every call requires X-Wever-Wallet-Proof bound to the exact JSON arguments and this API path, including paid calls. The tool fee is separate from the payout amount and grants no disbursement authority. The operator manually funds the dedicated hot payout wallet; Ledger keys never reach the server, and sweep and reclaim remain manual. No private keys are accepted. A disbursement primitive, not a bank: no yield, conversion or fiat. External custody and security audit required before large-value use. Company-authored; usage remains explicitly labeled. Read /api/payout-contract for exact schemas and authorization flow.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_purchase-statusRead your own purchase quote, recorded human approval, fulfillment and completion state.Reads text others writeNo findings
Why it is in this group
- The name "wever_purchase-status" says "status": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Read your own purchase quote, recorded human approval, fulfillment and completion state. Requires the same configured X-Wever-Agent-Key. No wallet or signing instructions are returned.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_quote-compareDiligenceOps: submit two to 20 provider quotes for the same job, trip leg or purchase (price, fees, tax, what is included or excluded, cancellation and refund terms, timing, validi…No findings
Why it is in this group
- The name "wever_quote-compare" says "compare": a word for reading.
Description, as the server wrote it DiligenceOps: submit two to 20 provider quotes for the same job, trip leg or purchase (price, fees, tax, what is included or excluded, cancellation and refund terms, timing, validity). The signed comparison gives exact totals and deltas to the lowest, a like-for-like total that prices in items another quote includes, the term differences that change the real price (separately priced items, exclusions, refundability, deposits at risk, expired quotes), and any quote missing terms the others state. POST /api/quote-compare. Reads only the material in the call: no page fetch, no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/diligence-checks-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_read-only-scope-checkAgent Access: submit an agent meant to investigate and change nothing, with every tool it can call: how each is described (read, query, write or admin), its operations, the free-fo…Reads private dataReads text others write2 findings
Why it is in this group
- The name "wever_read-only-scope-check" says "read": a word for reading.
What it can reach
- private-dataThe description's first sentence says "sql": it reads databases.
- untrusted-inputThe description's first sentence says "http": it reads web pages, which anyone can write.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...cript or a filter expression), what the credential behind it can do, the HTTP methods a re..."
- Info
The input schema at properties.tools.items.properties.query_access.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.tools.items.properties.query_access.description: "What the credential or role behind the tool can do."
Description, as the server wrote it Agent Access: submit an agent meant to investigate and change nothing, with every tool it can call: how each is described (read, query, write or admin), its operations, the free-form input it passes on (SQL, GraphQL, an HTTP request, a shell command, a script or a filter expression), what the credential behind it can do, the HTTP methods a request tool can send and any side effects. The signed report names every mutation path: write, delete, send, execute, configure, admin and money operations, SQL or GraphQL passed on under a credential that can write, request tools that can send POST, PUT, PATCH or DELETE, shell and script runners and stated side effects, with the writes hiding inside query-shaped tools listed apart, so the read-only boundary is real rather than described. Reachable actions against an authority grant are wever_authority-scope-check and a sandbox’s routes into production are wever_escape-path-check. Verdict mutation_reachable, boundary_unverified or read_only. POST /api/read-only-scope-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-access-contract. Human page: https://weverlabs.com/agent-access/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_sports-trip-statusRead the itinerary, current leg states, exact prices, approval records and signed receipts for your trip.Reads text others writeNo findings
Why it is in this group
- The name "wever_sports-trip-status" says "status": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Read the itinerary, current leg states, exact prices, approval records and signed receipts for your trip. POST /api/sports-trip-status. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the shared 10 signed-wallet free calls over rolling 30 days. X-Wever-Wallet-Proof binds the exact JSON arguments to this API path. API access fees are separate from travel spending. Per-leg principal approval and delegated authority govern spending. Conditional escrow has a 300 USDC per-leg cap. Duffel test mode uses a test token and simulated travel. Booking approval is separate from quote selection. Uncertain attempts are reconciled through trip-status. Read /api/sports-trips-contract for schemas, provider availability, receipt verification and approval terms.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
wever_status-evidence-checkDiligenceOps: submit a status page (text, or a public HTTPS url).Reads text others writeCan send data outNo findings
Why it is in this group
- The name "wever_status-evidence-check" says "status": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- untrusted-inputThe description's first sentence says "http": it reads web pages, which anyone can write.
- untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Claude Code: ask, because it is tagged outbound.
Description, as the server wrote it DiligenceOps: submit a status page (text, or a public HTTPS url). The signed report lists the uptime figures and whether each states a reporting window (past 90 days, this month, a date range), the incident and maintenance entries with their dates and status words, and the most recent entry with its date and days before the check. POST /api/status-evidence-check. Reads the supplied text, or one public HTTPS page read once with GET (no redirects, public DNS only, 512 KB): no account access, no merchant contact and no money movement. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. The report matches words, figures, dates and named items between the submitted texts; it does not interpret meaning beyond that matching. Contract: /api/diligence-evidence-contract. Human page: https://weverlabs.com/diligence-checks/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_thesis-compareThesis and mandate: compare one market across Kalshi, Polymarket and Novig.Reads text others writeNo findings
Why it is in this group
- The name "wever_thesis-compare" says "compare": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Thesis and mandate: compare one market across Kalshi, Polymarket and Novig. Input market: a topic such as Jaguars Ravens, a Kalshi ticker or a Kalshi or Polymarket market URL. For each outcome, returns each venue’s bid, ask and midpoint where it quotes, the thesis fair value weighted across quoted venues, the midpoint spread between every pair of venues and each venue’s gap to fair value, in points on a $1 payout basis. A venue that does not carry the market, or lists it without a usable quote, is named with the reason. Several matching games return candidates without charge. POST /api/thesis-compare. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/thesis-mandate-contract. Human page: https://weverlabs.com/thesis-mandate/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_treasury-balanceRead a wallet's native USDC balance and ETH balance for gas visibility on Base mainnet.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "read": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Read a wallet's native USDC balance and ETH balance for gas visibility on Base mainnet. wallet_address must use the exact EIP-55 checksum. Reads public chain data only. Treasury arguments never accept private keys, signatures or transaction instructions; these tools cannot sign, submit transactions, transfer funds or spend. POST /api/treasury-balance. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. API access billing is separate from wallet visibility: the existing access layer processes caller-supplied payment and wallet-proof headers only. X-Wever-Wallet-Proof must bind to the exact JSON arguments and this API path, not /mcp. The inspected wallet need not be the access payer. Read /api/treasury-contract and the backend 402 body for schemas, coverage and billing terms.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_treasury-flowsRead recent native USDC transfers into and out of a wallet on Base mainnet, including amounts, timestamps and transaction hashes.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "read": a word for reading.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Read recent native USDC transfers into and out of a wallet on Base mainnet, including amounts, timestamps and transaction hashes. Optional limit and UTC time window bound the returned history. wallet_address must use the exact EIP-55 checksum. Reads public chain data only. Treasury arguments never accept private keys, signatures or transaction instructions; these tools cannot sign, submit transactions, transfer funds or spend. POST /api/treasury-flows. Paid URL: 0.10 USDC per call on Base mainnet using x402 exact, under the existing shared 10 signed-wallet free calls over rolling 30 days. API access billing is separate from wallet visibility: the existing access layer processes caller-supplied payment and wallet-proof headers only. X-Wever-Wallet-Proof must bind to the exact JSON arguments and this API path, not /mcp. The inspected wallet need not be the access payer. Read /api/treasury-contract and the backend 402 body for schemas, coverage and billing terms.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
wever_watch-listWever Watch: list the calling wallet’s watches with state (pending_activation, active, fired, expired or cancelled), checks run, last observation and fire count.No findings
Why it is in this group
- The name "wever_watch-list" says "list": a word for reading.
Description, as the server wrote it Wever Watch: list the calling wallet’s watches with state (pending_activation, active, fired, expired or cancelled), checks run, last observation and fire count. Optional state filter. POST /api/watch-list. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/watch-contract. Human page: https://weverlabs.com/watch/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
wever_watch-statusWever Watch: one watch owned by the calling wallet in detail: terms, checks run, last observed value and evidence, every fire with its signed fire report and callback delivery outc…No findings
Why it is in this group
- The name "wever_watch-status" says "status": a word for reading.
Description, as the server wrote it Wever Watch: one watch owned by the calling wallet in detail: terms, checks run, last observed value and evidence, every fire with its signed fire report and callback delivery outcome. POST /api/watch-status. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Every call carries X-Wever-Wallet-Proof bound to the exact JSON arguments and the tool route; a paid call adds Payment-Signature from the same wallet. Returns a report with an Ed25519 signed receipt. Contract: /api/watch-contract. Human page: https://weverlabs.com/watch/.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Exfil surface
Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.
13992 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 3 private-data tools breaks every one.
- Reads private data3
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write106
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out44
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
- wever_agent-budget-guard
- wever_alert-route
- wever_api-change-check
- wever_boutique-confirm-delivery
- wever_budget-controls-check
- wever_changelog-check
- wever_consent-record-check
- wever_consent-revocation-check
- wever_consent-scope-check
- wever_consent-transfer-check
- wever_coordination-escalation-check
- wever_data-evidence-check
- and 32 more
{
"permissions": {
"deny": [
"mcp__wever-labs-products__wever_directory-status",
"mcp__wever-labs-products__wever_observability-log-check",
"mcp__wever-labs-products__wever_read-only-scope-check"
]
}
}Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.
Remote posture
What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.
Looked at Nov 20, 2026, 18:57 UTC, unchanged since Nov 20, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 200
- tools/list served without credentials
- no
- Tools that change things, listed without credentials
- 0
- Protocol version
- not observed
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- https://weverlabs.com/.well-known/oauth-protected-resource/mcp
- Authorization servers
- not observed
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- not observed
- Client ID metadata documents
- not observed
The probe stopped: The response is larger than 2097152 bytes.
Posture changes
- Nov 20, 2026
- tools/list served without credentials: yes to no
- Tools that change things, listed without credentials: 419 to 0
- Protocol version: 2025-06-18 to none
- Nov 13, 2026
- Tools that change things, listed without credentials: 157 to 419
- RFC 9728 protected resource metadata: none to https://weverlabs.com/.well-known/oauth-protected-resource/mcp
- Nov 6, 2026
- Tools that change things, listed without credentials: 65 to 157
- Oct 29, 2026
- Tools that change things, listed without credentials: 49 to 65
Findings
Matches of fixed checks, with the text they matched.
The description contains a 42-character base64 string.
What this check looks forMatched text
description: "...and API-fee wallet: 0x69255FeaC86268C0D0bD0CBd128da7B9aB44b1c4. Principal funding..."
The input schema at properties.terms.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.terms.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
The input schema at properties.funding_payment.anyOf[1].properties.accepted.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.funding_payment.anyOf[1].properties.accepted.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
The input schema at properties.funding_payment.oneOf[1].properties.accepted.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.funding_payment.oneOf[1].properties.accepted.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
The input schema at properties.mandate.properties.context.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.mandate.properties.context.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
The input schema at properties.mandate.properties.terms.properties.asset.const contains a 42-character base64 string.
What this check looks forMatched text
inputSchema.properties.mandate.properties.terms.properties.asset.const: "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913"
Show 349 more findingsShow fewer
The server marks "wever_dispute-resolve" destructive (destructiveHint: true), and its name says "resolve", a word for reading.
What this check looks forMatched text
annotations: {"readOnlyHint":false,"destructiveHint":true,"idempotentHint":true,"openWorldHint":true}; The name "wever_dispute-resolve" says "resolve": a word for reading.
The description contains "all other tools", wording about other tools or servers.
What this check looks forMatched text
description: "...e records. Real escrow/dispute caps and all other tools are unchanged."
The description contains "all other tools", wording about other tools or servers.
What this check looks forMatched text
description: "...e records. Real escrow/dispute caps and all other tools are unchanged."
The server marks "wever_second-date-resolve" as not read-only (readOnlyHint: false), and its words say it reads.
What this check looks forMatched text
annotations: {"readOnlyHint":false,"destructiveHint":false,"idempotentHint":true,"openWorldHint":true}; The name "wever_second-date-resolve" says "resolve": a word for reading.
The description names a tool "close_mailbox" that this server does not list.
What this check looks forMatched text
description: "...credential; use its inbox, get_task and close_mailbox tools. All agent steps use MCP; the server ha..."
The server marks "wever_policy-fit-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.amount) and a currency, so it moves money.
The server marks "wever_refund-eligibility-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_refund-eligibility-check" says "refund": a word that usually means money moves, which cannot be taken back.
The server marks "wever_schedule-list" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_schedule-list" says "schedule": a word for a change.
The server marks "wever_schedule-status" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_schedule-status" says "schedule": a word for a change.
The server marks "wever_payment-policy-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-policy-check" says "payment": a word that usually means money moves, which cannot be taken back.
The server marks "wever_accessibility-evidence-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input takes a statement or command to run (input.statement), which can change anything it can reach.
The server marks "wever_service-terms-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.exposure.amount) and a currency, so it moves money.
The server marks "wever_payment-instruction-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-instruction-check" says "payment": a word that usually means money moves, which cannot be taken back.
The server marks "wever_invoice-payment-match" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_invoice-payment-match" says "payment": a word that usually means money moves, which cannot be taken back.
The server marks "wever_expense-close-pack" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_expense-close-pack" says "close": a word for a change.
The server marks "wever_payout-readiness-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payout-readiness-check" says "payout": a word that usually means money moves, which cannot be taken back.
The server marks "wever_procurement-decision-brief" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.budget.amount) and a currency, so it moves money.
The server marks "wever_payment-rail-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-rail-check" says "payment": a word that usually means money moves, which cannot be taken back.
The server marks "wever_payment-recovery-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payment-recovery-check" says "payment": a word that usually means money moves, which cannot be taken back.
The server marks "wever_checkout-clarity-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_checkout-clarity-check" says "checkout": a word for a change.
The server marks "wever_payout-change-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_payout-change-check" says "payout": a word that usually means money moves, which cannot be taken back.
The server marks "wever_run-record-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_run-record-check" says "run": a word for a change.
The server marks "wever_money-fit-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "irreversible".
The server marks "wever_kill-switch-check" read-only (readOnlyHint: true), and its own name, description or input give the class "destructive".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_kill-switch-check" says "kill": a word for removing or ending something.
The server marks "wever_kill-switch-check" not destructive (destructiveHint: false), and its own name, description or input give the class "destructive".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_kill-switch-check" says "kill": a word for removing or ending something.
The server marks "wever_refund-rule-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_refund-rule-check" says "refund": a word that usually means money moves, which cannot be taken back.
The server marks "wever_budget-controls-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.recipients): what it sends cannot be taken back.
The server marks "wever_double-charge-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_double-charge-check" says "charge": a word that usually means money moves, which cannot be taken back.
The server marks "wever_refund-path-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_refund-path-check" says "refund": a word that usually means money moves, which cannot be taken back.
The server marks "wever_purchase-record-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_purchase-record-check" says "purchase": a word that usually means money moves, which cannot be taken back.
The server marks "wever_context-transfer-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_context-transfer-check" says "transfer": a word for a change that another call does not undo.
The server marks "wever_rollback-path-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "cannot be undone".
The server marks "wever_negotiation-mandate-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.escalation.to): what it sends cannot be taken back.
The server marks "wever_negotiation-settlement-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.settlement.amount) and a currency, so it moves money.
The server marks "wever_reputation-claim-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.claim.window.to): what it sends cannot be taken back.
The server marks "wever_reputation-report-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.window.to): what it sends cannot be taken back.
The server marks "wever_observability-metric-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.period.to): what it sends cannot be taken back.
The server marks "wever_observability-report-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.summary.period.to): what it sends cannot be taken back.
The server marks "wever_knowledge-update-check" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_knowledge-update-check" says "update": a word for a change.
The server marks "wever_coordination-escalation-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.path.recipient): what it sends cannot be taken back.
The server marks "wever_migration-return-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "cannot be undone".
The server marks "wever_feedback-request-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_feedback-request-check" says "request": a word for a change.
The server marks "wever_commitment-capture-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_commitment-capture-check" says "capture": a word that usually means money moves, which cannot be taken back.
The server marks "wever_learning-capture-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_learning-capture-check" says "capture": a word that usually means money moves, which cannot be taken back.
The server marks "wever_learning-transfer-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_learning-transfer-check" says "transfer": a word for a change that another call does not undo.
The server marks "wever_vendor-refund-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_vendor-refund-check" says "refund": a word that usually means money moves, which cannot be taken back.
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The server marks "wever_consent-scope-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.grant.recipients): what it sends cannot be taken back.
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The server marks "wever_consent-revocation-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.grant.recipients): what it sends cannot be taken back.
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The server marks "wever_consent-transfer-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_consent-transfer-check" says "transfer": a word for a change that another call does not undo.
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The description names a tool "wever_mandate" that this server does not list.
What this check looks forMatched text
description: "...t is not authority), not a mandate (the wever_mandate tools, wever_mandate-issue, wever_mandate-ame..."
The server marks "wever_consent-record-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.entry.recipients): what it sends cannot be taken back.
The server marks "wever_risk-reversibility-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The description says "cannot be undone".
The server marks "wever_onboarding-limits-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.handoff.to): what it sends cannot be taken back.
The server marks "wever_onboarding-readiness-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.escalation.to): what it sends cannot be taken back.
The server marks "wever_energy-demand-charge-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The name "wever_energy-demand-charge-check" says "charge": a word that usually means money moves, which cannot be taken back.
The server marks "wever_tokenized-cost-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.purchase.amount) and a currency, so it moves money.
The server marks "wever_role-scope-boundary-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.refusal.to): what it sends cannot be taken back.
The server marks "wever_role-handoff-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input addresses people (input.handoff.to): what it sends cannot be taken back.
The server marks "wever_travel-cancellation-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.paid.amount) and a currency, so it moves money.
The server marks "wever_insurance-limits-check" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
What this check looks forMatched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":false,"openWorldHint":false}; The input carries an amount (input.exposure.amount) and a currency, so it moves money.
The description mentions "Credentials".
What this check looks forMatched text
description: "...s and all other modes have no x402 fee. Credentials and grants remain required for their ex..."
The input schema at description mentions "Credentials".
What this check looks forMatched text
inputSchema.description: "...delegated-authority v2 request shapes. Credentials and issuance idempotency keys are HTTP..."
The input schema at oneOf[3].description mentions "credential".
What this check looks forMatched text
inputSchema.oneOf[3].description: "Public and read-only: no credential, fee or write. Send the key the mandate..."
The input schema at $defs.mandateInput.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.mandateInput.properties.usage.description: "...age and attribution come from the agent credential."
The input schema at $defs.mandateClaims.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.mandateClaims.properties.usage.description: "...age and attribution come from the agent credential."
The input schema at $defs.actionGrantClaims.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.actionGrantClaims.properties.usage.description: "...age and attribution come from the agent credential."
The input schema at $defs.signedMandate.description mentions "credentials".
What this check looks forMatched text
inputSchema.$defs.signedMandate.description: "...re, active time window, durable record, credentials and claim bindings. Successful runtime..."
The input schema at $defs.signedActionGrant.description mentions "credentials".
What this check looks forMatched text
inputSchema.$defs.signedActionGrant.description: "...re, active time window, durable record, credentials and claim bindings. Successful runtime..."
The input schema at $defs.grantRequest.properties.usage.description mentions "credential".
What this check looks forMatched text
inputSchema.$defs.grantRequest.properties.usage.description: "...age and attribution come from the agent credential."
The description is 2,263 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...kout intent and grants no tool access, execution authority or spending authority."
The description mentions "credentials".
What this check looks forMatched text
description: "...i/agent-paid-workflow. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...nified-agent-checkout. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...diligenceops-paid-run. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: ".../api/receipt-passport. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...return-package-viewer. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...ent-allowance-console. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...llback-health-monitor. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...t-work-order-exchange. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...pi/agent-budget-guard. Existing product credentials, signed mandates, and single-use action..."
The input schema at properties.policy.properties.allowed_callback_origins.items.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.policy.properties.allowed_callback_origins.items.description: "...gin only, with no path, trailing slash, credentials, query or fragment. Matching uses the p..."
The input schema at properties.request.properties.callback_url.anyOf[0].description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.request.properties.callback_url.anyOf[0].description: "ASCII HTTPS URL without credentials, control characters, backslashes or fra..."
The description mentions "credentials".
What this check looks forMatched text
description: "...T /api/agent-sla-rail. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...agent-trust-scorecard. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...t-authority-inspector. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: ".../x402-payment-gateway. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...i/ap2-mandate-gateway. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...to-agent-handoff-pack. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...POST /api/proof-inbox. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: ".../api/rail-playground. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...api/agent-escrow-lite. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...pi/agent-invoice-rail. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...ent-permission-ledger. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...agent-work-order-rail. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...-decision-packet-rail. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: ".../agent-file-desk-rail. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...pi/mcp-diligence-rail. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...-daily-operator-brief. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...POST /api/proof-relay. Existing product credentials, signed mandates, and single-use action..."
The description mentions "credentials".
What this check looks forMatched text
description: "...s adapter forwards only caller-supplied credentials and grants no spending authority."
The input schema at properties.claim.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.claim.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The input schema at properties.references.items.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.references.items.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The input schema at properties.reason_uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.reason_uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The input schema at properties.claim.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.claim.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The input schema at properties.references.items.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.references.items.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The description mentions "credential".
What this check looks forMatched text
description: "...nd API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans a..."
The description mentions "credential".
What this check looks forMatched text
description: "...nd API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans a..."
The description mentions "credential".
What this check looks forMatched text
description: "...nd API path. Send your existing mailbox credential in X-Wever-Connect-Token. Both humans a..."
The input schema at oneOf[3].properties.claim.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.oneOf[3].properties.claim.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The input schema at oneOf[4].properties.references.items.properties.uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.oneOf[4].properties.references.items.properties.uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The input schema at oneOf[5].properties.decision.properties.reason_uri.description mentions "credentials".
What this check looks forMatched text
inputSchema.oneOf[5].properties.decision.properties.reason_uri.description: "...erence. No raw evidence, personal data, credentials, query strings, or fragments."
The title mentions "credential".
What this check looks forMatched text
title: "Issue a verified single credential"
The description mentions "credential".
What this check looks forMatched text
description: "...d over rolling 30 days. Issue the exact credential authorized by a human at https://weverl..."
The description mentions "credential".
What this check looks forMatched text
description: "...eturns body.mailbox.mailbox_id and body.credential.access_token once. Save the credential..."
The description is 2,179 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ng Connect registration quotas, private visibility and human approval gate apply."
The description mentions "credential".
What this check looks forMatched text
description: "...as anonymous access, payment required, credential required or unreachable; and the checke..."
The input schema at properties.fields.properties.key_custody.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.fields.properties.key_custody.description: "Who holds the signing keys and credentials, and any backup or recovery for them."
The description mentions "credentials".
What this check looks forMatched text
description: "...o server name, a URL that is not HTTPS, credentials in the address, an auth header that is..."
The description mentions "credentials".
What this check looks forMatched text
description: "...eceiver takes over. Input text, fields (credentials, runbooks, billing, domains, data_expor..."
The input schema at properties.fields.properties.credentials.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.fields.properties.credentials.description: "How credentials and key custody pass to the receiver."
The description mentions "credentials".
What this check looks forMatched text
description: "...nly, 5 seconds, a response size cap, no credentials) and never pays it. The signed report s..."
The input schema at properties.access.items.properties.via.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.access.items.properties.via.description: "The credential or role."
The title mentions "Credential".
What this check looks forMatched text
title: "Credential requirements check"
The description mentions "credentials".
What this check looks forMatched text
description: "...in order (each step with its words, the credentials, accounts or approvals it asks for, and..."
The input schema at properties.flow.items.properties.requires.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.flow.items.properties.requires.description: "Credentials, accounts or approvals first asked for..."
The input schema at properties.components.items.properties.kind.enum[5] mentions "credential".
What this check looks forMatched text
inputSchema.properties.components.items.properties.kind.enum[5]: "credential"
The description mentions "credentials".
What this check looks forMatched text
description: "...paration design: sandbox and production credentials (who holds each, whether production acc..."
The input schema at required[0] mentions "credentials".
What this check looks forMatched text
inputSchema.required[0]: "credentials"
The input schema at properties.credentials.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.credentials.description: "Credentials in either environment."
The input schema at properties.endpoints.items.properties.accepts.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.endpoints.items.properties.accepts.description: "Credentials it accepts; an empty list means none."
The description mentions "credential".
What this check looks forMatched text
description: "...and status) and the deployment, usage, credential or billing lists it should cover. The s..."
The input schema at properties.sources.items.properties.kind.enum[2] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.sources.items.properties.kind.enum[2]: "credentials"
The description mentions "credentials".
What this check looks forMatched text
description: "...dbox environment design for agents: the credentials it holds (where each is accepted and ho..."
The input schema at properties.credentials.items.properties.valid_in.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.valid_in.description: "Where the credential is accepted."
The input schema at properties.credentials.items.properties.form.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.form.description: "How the credential is made."
The input schema at properties.credentials.description mentions "Credentials".
What this check looks forMatched text
inputSchema.properties.credentials.description: "Credentials the sandbox holds or can read."
The description mentions "credential".
What this check looks forMatched text
description: "...cript or a filter expression), what the credential behind it can do, the HTTP methods a re..."
The input schema at properties.tools.items.properties.query_access.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.tools.items.properties.query_access.description: "What the credential or role behind the tool can do."
The description mentions "credential".
What this check looks forMatched text
description: "...ated (gateway, policy engine, platform, credential scope, tool code, database, system prom..."
The description is 2,075 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-purchases-contract. Human page: https://weverlabs.com/agent-purchases/."
The description mentions "credential".
What this check looks forMatched text
description: "...erchant, category, amount, time and the credential paid with) and the delegation behind it..."
The input schema at properties.receipt.properties.credential.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.receipt.properties.credential.description: "The credential the payment was made with."
The description mentions "credential".
What this check looks forMatched text
description: "...way, the platform, a wallet policy or a credential scope outside the agent, or a system pr..."
The input schema at properties.limits.items.properties.enforced_at.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.limits.items.properties.enforced_at.description: "...de, gateway, platform, wallet policy or credential scope outside the agent, or a system pr..."
The description mentions "credential".
What this check looks forMatched text
description: "...registry record, by its signature, its credential or mutual TLS), when (before the task m..."
The input schema at properties.routes.items.properties.verified_by.items.enum[2] mentions "credential".
What this check looks forMatched text
inputSchema.properties.routes.items.properties.verified_by.items.enum[2]: "credential"
The input schema at properties.routes.items.properties.verified_by.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.routes.items.properties.verified_by.description: "...gistry record, by its signature, by its credential or by mutual TLS."
The description mentions "credential".
What this check looks forMatched text
description: "...registry covers what deployment, usage, credential and billing lists show is wever_agent-r..."
The description mentions "credential".
What this check looks forMatched text
description: "..., the receiver’s own standing access, a credential several agents share or the sender’s ow..."
The input schema at properties.delegations.items.properties.grant.properties.via.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.delegations.items.properties.grant.properties.via.description: "..., the receiver’s own standing access, a credential several agents share, or the sender’s o..."
The description is 2,076 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/."
The description is 2,045 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-recovery-contract. Human page: https://weverlabs.com/agent-recovery/."
The description mentions "credentials".
What this check looks forMatched text
description: "...l, task history, preferences, contacts, credentials, payment details, health details, locat..."
The input schema at properties.categories.items.properties.kind.enum[4] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.categories.items.properties.kind.enum[4]: "credentials"
The input schema at properties.categories.items.properties.kind.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.categories.items.properties.kind.description: "...l, task history, preferences, contacts, credentials, payment details, health details, locat..."
The description is 2,162 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
The description is 2,040 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
The description is 2,248 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
The description is 2,211 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
The description is 2,210 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
The description is 2,109 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
The description is 2,058 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-memory-contract. Human page: https://weverlabs.com/agent-memory/."
The description is 2,120 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
The description mentions "credential".
What this check looks forMatched text
description: "...ainst what (a registry record, a signed credential, an attester, a prior working relations..."
The input schema at properties.identity.properties.method.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.identity.properties.method.description: "...ed against: a registry record, a signed credential, an attester, a prior working relations..."
The description is 2,020 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
The description is 2,180 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
The description is 2,085 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
The description is 2,053 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
The description is 2,217 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-delegation-contract. Human page: https://weverlabs.com/agent-delegation/."
The description is 2,183 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
The description is 2,277 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
The description is 2,191 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
The description is 2,076 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
The description is 2,051 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-evaluation-contract. Human page: https://weverlabs.com/agent-evaluation/."
The description is 2,049 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/."
The description mentions "credential".
What this check looks forMatched text
description: "...ts, its format (JSON, CSV, a verifiable credential, signed attestations, on chain, PDF, HT..."
The description is 2,152 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-reputation-contract. Human page: https://weverlabs.com/agent-reputation/."
The description is 2,144 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t-observability-contract. Human page: https://weverlabs.com/agent-observability/."
The description mentions "credentials".
What this check looks forMatched text
description: "...an with no stated basis, and secrets or credentials written into traces and logs (private k..."
The description is 2,172 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/."
The description is 2,035 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/."
The description mentions "credentials".
What this check looks forMatched text
description: "...s class (data, memory, tools, identity, credentials, funds, history, settings, integrations..."
The input schema at properties.assets.items.properties.class.enum[4] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.assets.items.properties.class.enum[4]: "credentials"
The input schema at properties.assets.items.properties.class.description mentions "credentials".
What this check looks forMatched text
inputSchema.properties.assets.items.properties.class.description: "Data, memory, tools, identity, credentials, funds, history, settings, integrations..."
The input schema at properties.inventory.items.properties.class.enum[4] mentions "credentials".
What this check looks forMatched text
inputSchema.properties.inventory.items.properties.class.enum[4]: "credentials"
The description is 2,002 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-migration-contract. Human page: https://weverlabs.com/agent-migration/."
The description is 2,274 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,484 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,600 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,470 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,386 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,481 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,557 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,533 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-feedback-contract. Human page: https://weverlabs.com/agent-feedback/."
The description is 2,843 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 2,971 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 2,941 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 2,739 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 2,765 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 2,946 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 2,601 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 3,007 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-uncertainty-contract. Human page: https://weverlabs.com/agent-uncertainty/."
The description is 2,953 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 2,828 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 2,866 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 2,652 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 3,011 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 3,085 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 2,732 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 2,848 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...agent-commitments-contract. Human page: https://weverlabs.com/agent-commitments/."
The description is 3,105 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 2,767 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 2,951 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 2,890 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 2,887 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 2,855 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 3,090 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 2,802 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-learning-contract. Human page: https://weverlabs.com/agent-learning/."
The description is 2,750 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 2,939 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 2,968 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 2,891 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 2,783 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 3,133 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 2,802 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 3,084 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-triage-contract. Human page: https://weverlabs.com/agent-triage/."
The description is 3,652 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 3,836 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 3,807 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 3,561 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 3,747 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 3,455 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 3,700 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 3,699 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-goals-contract. Human page: https://weverlabs.com/agent-goals/."
The description is 4,316 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 4,227 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 4,381 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 4,322 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 4,114 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 4,252 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 4,325 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 4,207 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-vendors-contract. Human page: https://weverlabs.com/agent-vendors/."
The description is 3,936 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 3,809 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 4,125 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 4,121 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 4,009 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 4,467 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 4,001 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 4,099 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...t: /api/agent-consent-contract. Human page: https://weverlabs.com/agent-consent/."
The description is 4,231 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description is 4,157 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description is 3,850 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description is 4,080 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description is 3,877 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description mentions "credential".
What this check looks forMatched text
description: "...a price, a person, an approval, data, a credential or other), whether the action fails or..."
The input schema at properties.dependencies.items.properties.kind.enum[5] mentions "credential".
What this check looks forMatched text
inputSchema.properties.dependencies.items.properties.kind.enum[5]: "credential"
The input schema at properties.dependencies.items.properties.kind.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.dependencies.items.properties.kind.description: "...a price, a person, an approval, data, a credential or other."
The description is 3,774 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description is 4,222 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description is 3,848 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...ontract: /api/agent-risk-contract. Human page: https://weverlabs.com/agent-risk/."
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
The description is 4,650 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The title mentions "credential".
What this check looks forMatched text
title: "Onboarding credential check"
The description mentions "credentials".
What this check looks forMatched text
description: "Agent Onboarding: submit the credentials a new agent was issued and the text its..."
The input schema at required[0] mentions "credentials".
What this check looks forMatched text
inputSchema.required[0]: "credentials"
The input schema at properties.credentials.items.properties.id.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.id.description: "The credential id, such as c1."
The input schema at properties.credentials.items.properties.name.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.items.properties.name.description: "The credential, such as the orders API key."
The input schema at properties.credentials.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.credentials.description: "Each credential the agent was issued; empty when none."
The input schema at properties.texts.items.properties.text.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.texts.items.properties.text.description: "The text as written, read for credential-shaped strings; the report names each b..."
The description is 4,718 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
The description is 4,549 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
The description is 4,430 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
The description is 4,623 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
The description is 4,277 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The description mentions "credential".
What this check looks forMatched text
description: "...figuration change or external write; no credential is issued, rotated or revoked, no permi..."
The description is 4,493 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The description mentions "credential".
What this check looks forMatched text
description: "...ers API read) and kind (a permission, a credential, a wallet, a limit, a system or other),..."
The input schema at properties.record.properties.granted.items.properties.kind.enum[1] mentions "credential".
What this check looks forMatched text
inputSchema.properties.record.properties.granted.items.properties.kind.enum[1]: "credential"
The input schema at properties.record.properties.granted.items.properties.kind.description mentions "credential".
What this check looks forMatched text
inputSchema.properties.record.properties.granted.items.properties.kind.description: "A permission, a credential, a wallet, a limit, a system or other."
The input schema at properties.live.items.properties.kind.enum[1] mentions "credential".
What this check looks forMatched text
inputSchema.properties.live.items.properties.kind.enum[1]: "credential"
The description is 4,178 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...i/agent-onboarding-contract. Human page: https://weverlabs.com/agent-onboarding/."
The description is 3,397 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,846 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,468 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,513 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,627 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,908 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,746 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,734 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-energy-contract. Human page: https://weverlabs.com/agent-energy/."
The description is 3,838 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 3,705 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 3,634 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 3,616 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 3,664 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 3,689 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 4,070 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 3,909 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-datasets-contract. Human page: https://weverlabs.com/agent-datasets/."
The description is 3,622 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,774 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,470 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,337 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,229 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,532 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,497 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,282 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...d-purchase-contract. Human page: https://weverlabs.com/tokenized-asset-purchase/."
The description is 3,069 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 3,427 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 3,523 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 3,460 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 3,421 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 2,991 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 3,221 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 3,514 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...tract: /api/agent-roles-contract. Human page: https://weverlabs.com/agent-roles/."
The description is 3,479 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,460 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,518 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,439 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,277 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,443 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,481 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,550 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: ".../api/agent-capacity-contract. Human page: https://weverlabs.com/agent-capacity/."
The description is 3,618 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,536 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,632 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,634 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,637 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,799 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,484 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,766 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...act: /api/agent-travel-contract. Human page: https://weverlabs.com/agent-travel/."
The description is 3,796 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
The description is 3,750 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
The description is 3,811 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
The description is 3,805 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
The description is 4,031 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
The description is 4,131 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
The description is 3,837 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
The description is 3,846 characters long, over the 2,000 this check looks for.
What this check looks forMatched text
description ends: "...api/agent-insurance-contract. Human page: https://weverlabs.com/agent-insurance/."
History
Changes to its tool definitions and to its remote endpoint, newest first.
PostureRemote posture changed. tools/list served without credentials: yes -> no; Tools that change things, listed without credentials: 419 -> 0; Protocol version: 2025-06-18 -> none.View diff
- tools/list served without credentials: yes to no
- Tools that change things, listed without credentials: 419 to 0
- Protocol version: 2025-06-18 to none
Tools8 added.+8 addedView diff
- Added:wever_insurance-certificate-checkChanges things
- Added:wever_insurance-claims-deadline-checkChanges things
- Added:wever_insurance-coverage-checkChanges things
- Added:wever_insurance-deductible-checkChanges things
- Added:wever_insurance-exclusions-checkChanges things
- Added:wever_insurance-limits-checkCan't be undone
- Added:wever_insurance-notice-checkChanges things
- Added:wever_insurance-renewal-checkChanges things
Tools8 added.+8 addedView diff
- Added:wever_travel-ancillary-checkChanges things
- Added:wever_travel-authorization-checkChanges things
- Added:wever_travel-cancellation-checkCan't be undone
- Added:wever_travel-change-checkChanges things
- Added:wever_travel-disruption-checkChanges things
- Added:wever_travel-fare-checkChanges things
- Added:wever_travel-itinerary-checkChanges things
- Added:wever_travel-passenger-checkChanges things
Tools8 added.+8 addedView diff
- Added:wever_capacity-backlog-checkChanges things
- Added:wever_capacity-burn-checkChanges things
- Added:wever_capacity-concurrency-checkChanges things
- Added:wever_capacity-headroom-checkChanges things
- Added:wever_capacity-queue-checkChanges things
- Added:wever_capacity-scaling-checkChanges things
- Added:wever_capacity-throttle-checkChanges things
- Added:wever_capacity-workload-checkChanges things
Tools8 added.+8 addedView diff
- Added:wever_role-definition-checkChanges things
- Added:wever_role-delegation-checkChanges things
- Added:wever_role-handoff-checkCan't be undone
- Added:wever_role-owner-checkChanges things
- Added:wever_role-permission-match-checkChanges things
- Added:wever_role-review-checkChanges things
- Added:wever_role-scope-boundary-checkCan't be undone
- Added:wever_role-separation-checkChanges things
Tools8 added.+8 addedView diff
- Added:wever_tokenized-backing-checkChanges things
- Added:wever_tokenized-cost-checkCan't be undone
- Added:wever_tokenized-custody-checkChanges things
- Added:wever_tokenized-platform-terms-checkChanges things
- Added:wever_tokenized-quote-checkChanges things
- Added:wever_tokenized-redemption-checkChanges things
- Added:wever_tokenized-settlement-checkChanges things
- Added:wever_tokenized-transferability-checkChanges things
Tools8 added.+8 addedView diff
- Added:wever_dataset-coverage-checkChanges things
- Added:wever_dataset-freshness-checkChanges things
- Added:wever_dataset-license-checkChanges things
- Added:wever_dataset-lineage-checkChanges things
- Added:wever_dataset-privacy-checkChanges things
- Added:wever_dataset-quality-checkChanges things
- Added:wever_dataset-schema-drift-checkChanges things
- Added:wever_dataset-synthetic-disclosure-checkChanges things
Tools8 added.+8 addedView diff
- Added:wever_energy-bill-checkChanges things
- Added:wever_energy-carbon-claim-checkChanges things
- Added:wever_energy-contract-checkChanges things
- Added:wever_energy-demand-charge-checkCan't be undone
- Added:wever_energy-efficiency-claim-checkChanges things
- Added:wever_energy-ev-charging-checkChanges things
- Added:wever_energy-solar-quote-checkChanges things
- Added:wever_energy-tariff-checkChanges things
PostureRemote posture changed. Tools that change things, listed without credentials: 157 -> 419; RFC 9728 protected resource metadata: none -> https://weverlabs.com/.well-known/oauth-protected-resource/mcp.View diff
- Tools that change things, listed without credentials: 157 to 419
- RFC 9728 protected resource metadata: none to https://weverlabs.com/.well-known/oauth-protected-resource/mcp
Tools8 added.+8 addedView diff
- Added:wever_onboarding-access-checkChanges things
- Added:wever_onboarding-credential-checkChanges things
- Added:wever_onboarding-identity-checkChanges things
- Added:wever_onboarding-limits-checkCan't be undone
- Added:wever_onboarding-permissions-checkChanges things
- Added:wever_onboarding-readiness-checkCan't be undone
- Added:wever_onboarding-record-checkChanges things
- Added:wever_onboarding-wallet-checkChanges things
Install
Pinned to version 0.8.2-registry.44. Values in angle brackets are yours to fill in.
claude mcp add --transport http wever-labs-products https://weverlabs.com/mcpclaude mcp add --transport http wever-labs-products https://weverlabs.com/mcpRun in a terminal.
Claude Code permissions
Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [
"mcp__wever-labs-products__wever_atlas-asset-search",
"mcp__wever-labs-products__wever_treasury-balance",
"mcp__wever-labs-products__wever_treasury-flows",
"mcp__wever-labs-products__wever_payout-status",
"mcp__wever-labs-products__wever_dispute-status",
"mcp__wever-labs-products__wever_boutique-list-products",
"mcp__wever-labs-products__wever_boutique-product-detail",
"mcp__wever-labs-products__wever_atlas-compare",
"mcp__wever-labs-products__wever_atlas-rights-gap",
"mcp__wever-labs-products__wever_purchase-status",
"mcp__wever-labs-products__wever_sports-trip-status",
"mcp__wever-labs-products__boutique-coupon",
"mcp__wever-labs-products__wever_mandate-status",
"mcp__wever-labs-products__wever_watch-list",
"mcp__wever-labs-products__wever_watch-status",
"mcp__wever-labs-products__wever_thesis-compare",
"mcp__wever-labs-products__wever_contract-diff-check",
"mcp__wever-labs-products__wever_quote-compare",
"mcp__wever-labs-products__wever_digest-status",
"mcp__wever-labs-products__wever_read-only-scope-check",
"mcp__wever-labs-products__wever_observability-log-check"
],
"ask": [
"mcp__wever-labs-products__wever_agent-paid-workflow",
"mcp__wever-labs-products__wever_unified-agent-checkout",
"mcp__wever-labs-products__wever_diligenceops-paid-run",
"mcp__wever-labs-products__wever_agent-allowance-console",
"mcp__wever-labs-products__wever_agent-work-order-exchange",
"mcp__wever-labs-products__wever_agent-budget-guard",
"mcp__wever-labs-products__wever_payment-authority-inspector",
"mcp__wever-labs-products__wever_x402-payment-gateway",
"mcp__wever-labs-products__wever_agent-work-order-rail",
"mcp__wever-labs-products__wever_mcp-diligence-rail",
"mcp__wever-labs-products__wever_agent-daily-operator-brief",
"mcp__wever-labs-products__wever_proof-relay",
"mcp__wever-labs-products__wever_atlas-asset-passport",
"mcp__wever-labs-products__wever_boutique-create-order",
"mcp__wever-labs-products__wever_boutique-record-shipment",
"mcp__wever-labs-products__wever_boutique-confirm-delivery",
"mcp__wever-labs-products__wever_boutique-real-money-acceptance",
"mcp__wever-labs-products__wever_sports-plan-trip",
"mcp__wever-labs-products__wever_matchmaking-opt-in-profile",
"mcp__wever-labs-products__wever_matchmaking-propose-intro",
"mcp__wever-labs-products__wever_matchmaking-approve-or-decline-intro",
"mcp__wever-labs-products__wever_post-job",
"mcp__wever-labs-products__wever_submit-work",
"mcp__wever-labs-products__edge-alert",
"mcp__wever-labs-products__fair-value",
"mcp__wever-labs-products__wever_single-credential-issue",
"mcp__wever-labs-products__wever_second-date-market-open",
"mcp__wever-labs-products__wever_second-date-bet",
"mcp__wever-labs-products__wever_second-date-resolve",
"mcp__wever-labs-products__wever_mandate-issue",
"mcp__wever-labs-products__wever_mandate-amend",
"mcp__wever-labs-products__wever_matchmaking-register-mailbox",
"mcp__wever-labs-products__wever_data-evidence-check",
"mcp__wever-labs-products__wever_watch-create",
"mcp__wever-labs-products__wever_receipt-verify",
"mcp__wever-labs-products__wever_escrow-quote",
"mcp__wever-labs-products__wever_invoice-check",
"mcp__wever-labs-products__wever_terms-evidence-check",
"mcp__wever-labs-products__wever_counterparty-brief",
"mcp__wever-labs-products__wever_wallet-activity-brief",
"mcp__wever-labs-products__wever_fair-value-brief",
"mcp__wever-labs-products__wever_deadline-check",
"mcp__wever-labs-products__wever_listing-check",
"mcp__wever-labs-products__wever_market-data-check",
"mcp__wever-labs-products__wever_mandate-draft",
"mcp__wever-labs-products__wever_claim-evidence-check",
"mcp__wever-labs-products__wever_policy-fit-check",
"mcp__wever-labs-products__wever_job-brief",
"mcp__wever-labs-products__wever_deliverable-check",
"mcp__wever-labs-products__wever_refund-eligibility-check",
"mcp__wever-labs-products__wever_subscription-audit",
"mcp__wever-labs-products__wever_schedule-create",
"mcp__wever-labs-products__wever_schedule-list",
"mcp__wever-labs-products__wever_schedule-status",
"mcp__wever-labs-products__wever_digest-create",
"mcp__wever-labs-products__wever_alert-route",
"mcp__wever-labs-products__wever_changelog-check",
"mcp__wever-labs-products__wever_pricing-evidence-check",
"mcp__wever-labs-products__wever_status-evidence-check",
"mcp__wever-labs-products__wever_docs-freshness-check",
"mcp__wever-labs-products__wever_license-evidence-check",
"mcp__wever-labs-products__wever_sla-evidence-check",
"mcp__wever-labs-products__wever_benchmark-evidence-check",
"mcp__wever-labs-products__wever_api-change-check",
"mcp__wever-labs-products__wever_atlas-issuer-brief",
"mcp__wever-labs-products__wever_atlas-rights-history",
"mcp__wever-labs-products__wever_atlas-redemption-check",
"mcp__wever-labs-products__wever_atlas-nav-evidence",
"mcp__wever-labs-products__wever_warranty-check",
"mcp__wever-labs-products__wever_shipping-check",
"mcp__wever-labs-products__wever_privacy-evidence-check",
"mcp__wever-labs-products__wever_certification-evidence-check",
"mcp__wever-labs-products__wever_testimonial-evidence-check",
"mcp__wever-labs-products__wever_fee-evidence-check",
"mcp__wever-labs-products__wever_translation-evidence-check",
"mcp__wever-labs-products__wever_environmental-evidence-check",
"mcp__wever-labs-products__wever_contract-brief",
"mcp__wever-labs-products__wever_policy-brief",
"mcp__wever-labs-products__wever_atlas-stage-brief",
"mcp__wever-labs-products__wever_atlas-market-brief",
"mcp__wever-labs-products__wever_sla-credit-check",
"mcp__wever-labs-products__wever_incident-evidence-check",
"mcp__wever-labs-products__wever_subprocessor-evidence-check",
"mcp__wever-labs-products__wever_guarantee-evidence-check",
"mcp__wever-labs-products__wever_pricing-tier-check",
"mcp__wever-labs-products__wever_export-evidence-check",
"mcp__wever-labs-products__wever_invoice-brief",
"mcp__wever-labs-products__wever_meeting-brief",
"mcp__wever-labs-products__wever_escrow-brief",
"mcp__wever-labs-products__wever_support-brief",
"mcp__wever-labs-products__wever_endpoint-reliability-check",
"mcp__wever-labs-products__wever_directory-submit",
"mcp__wever-labs-products__wever_directory-status",
"mcp__wever-labs-products__wever_wallet-statement",
"mcp__wever-labs-products__wever_payment-policy-check",
"mcp__wever-labs-products__wever_delegation-check",
"mcp__wever-labs-products__wever_agent-identity-check",
"mcp__wever-labs-products__wever_continuity-check",
"mcp__wever-labs-products__wever_security-disclosure-check",
"mcp__wever-labs-products__wever_data-retention-check",
"mcp__wever-labs-products__wever_accessibility-evidence-check",
"mcp__wever-labs-products__wever_ai-disclosure-check",
"mcp__wever-labs-products__wever_agent-card-check",
"mcp__wever-labs-products__wever_mcp-server-check",
"mcp__wever-labs-products__wever_succession-brief",
"mcp__wever-labs-products__wever_handover-check",
"mcp__wever-labs-products__wever_recovery-readiness-check",
"mcp__wever-labs-products__wever_bazaar-declaration-check",
"mcp__wever-labs-products__wever_directory-listing-check",
"mcp__wever-labs-products__wever_endpoint-honesty-check",
"mcp__wever-labs-products__wever_spending-policy-check",
"mcp__wever-labs-products__wever_approval-chain-check",
"mcp__wever-labs-products__wever_service-terms-check",
"mcp__wever-labs-products__wever_incident-plan-check",
"mcp__wever-labs-products__wever_insurance-evidence-check",
"mcp__wever-labs-products__wever_audit-evidence-check",
"mcp__wever-labs-products__wever_dpa-evidence-check",
"mcp__wever-labs-products__wever_nda-evidence-check",
"mcp__wever-labs-products__wever_payment-instruction-check",
"mcp__wever-labs-products__wever_invoice-payment-match",
"mcp__wever-labs-products__wever_settlement-reconciliation-check",
"mcp__wever-labs-products__wever_expense-close-pack",
"mcp__wever-labs-products__wever_payout-readiness-check",
"mcp__wever-labs-products__wever_counterparty-onboarding-check",
"mcp__wever-labs-products__wever_procurement-decision-brief",
"mcp__wever-labs-products__wever_renewal-notice-check",
"mcp__wever-labs-products__wever_spending-anomaly-check",
"mcp__wever-labs-products__wever_budget-forecast-brief",
"mcp__wever-labs-products__wever_payment-rail-check",
"mcp__wever-labs-products__wever_exchange-quote-check",
"mcp__wever-labs-products__wever_treasury-policy-check",
"mcp__wever-labs-products__wever_invoice-aging-brief",
"mcp__wever-labs-products__wever_fee-reconciliation-check",
"mcp__wever-labs-products__wever_spend-category-brief",
"mcp__wever-labs-products__wever_job-cost-brief",
"mcp__wever-labs-products__wever_job-receipt-pack",
"mcp__wever-labs-products__wever_policy-replay-check",
"mcp__wever-labs-products__wever_duplicate-order-check",
"mcp__wever-labs-products__wever_payment-recovery-check",
"mcp__wever-labs-products__wever_terms-change-check",
"mcp__wever-labs-products__wever_recommendation-disclosure-check",
"mcp__wever-labs-products__wever_job-handoff-pack",
"mcp__wever-labs-products__wever_checkout-clarity-check",
"mcp__wever-labs-products__wever_approval-version-check",
"mcp__wever-labs-products__wever_trace-outcome-brief",
"mcp__wever-labs-products__wever_answer-freshness-check",
"mcp__wever-labs-products__wever_context-sharing-check",
"mcp__wever-labs-products__wever_agent-attention-brief",
"mcp__wever-labs-products__wever_payout-change-check",
"mcp__wever-labs-products__wever_paid-access-check",
"mcp__wever-labs-products__wever_approval-replay-check",
"mcp__wever-labs-products__wever_spec-drift-check",
"mcp__wever-labs-products__wever_action-threshold-check",
"mcp__wever-labs-products__wever_ownership-map-brief",
"mcp__wever-labs-products__wever_catalog-listing-check",
"mcp__wever-labs-products__wever_policy-precedence-check",
"mcp__wever-labs-products__wever_settled-value-check",
"mcp__wever-labs-products__wever_memory-design-check",
"mcp__wever-labs-products__wever_run-record-check",
"mcp__wever-labs-products__wever_money-fit-check",
"mcp__wever-labs-products__wever_adoption-readiness-check",
"mcp__wever-labs-products__wever_authority-scope-check",
"mcp__wever-labs-products__wever_change-proof-check",
"mcp__wever-labs-products__wever_credential-requirements-check",
"mcp__wever-labs-products__wever_retry-safety-check",
"mcp__wever-labs-products__wever_router-disclosure-check",
"mcp__wever-labs-products__wever_firewall-order-check",
"mcp__wever-labs-products__wever_escape-path-check",
"mcp__wever-labs-products__wever_escalation-threshold-check",
"mcp__wever-labs-products__wever_agent-registry-check",
"mcp__wever-labs-products__wever_deterministic-core-check",
"mcp__wever-labs-products__wever_approval-durability-check",
"mcp__wever-labs-products__wever_cache-staleness-check",
"mcp__wever-labs-products__wever_refund-rule-check",
"mcp__wever-labs-products__wever_actor-attribution-check",
"mcp__wever-labs-products__wever_approval-policy-check",
"mcp__wever-labs-products__wever_volume-quality-check",
"mcp__wever-labs-products__wever_money-choice-check",
"mcp__wever-labs-products__wever_memory-separation-check",
"mcp__wever-labs-products__wever_identity-portability-check",
"mcp__wever-labs-products__wever_gateway-coverage-check",
"mcp__wever-labs-products__wever_sandbox-isolation-check",
"mcp__wever-labs-products__wever_exposure-cap-check",
"mcp__wever-labs-products__wever_discovery-coverage-check",
"mcp__wever-labs-products__wever_liability-terms-check",
"mcp__wever-labs-products__wever_approval-record-check",
"mcp__wever-labs-products__wever_budget-controls-check",
"mcp__wever-labs-products__wever_incident-ownership-check",
"mcp__wever-labs-products__wever_terms-match-check",
"mcp__wever-labs-products__wever_tool-exposure-check",
"mcp__wever-labs-products__wever_gateway-rules-check",
"mcp__wever-labs-products__wever_transaction-gate-check",
"mcp__wever-labs-products__wever_public-surface-check",
"mcp__wever-labs-products__wever_enforcement-layer-check",
"mcp__wever-labs-products__wever_external-listing-check",
"mcp__wever-labs-products__wever_settlement-proof-check",
"mcp__wever-labs-products__wever_provider-readiness-check",
"mcp__wever-labs-products__wever_rehearsal-check",
"mcp__wever-labs-products__wever_price-truth-check",
"mcp__wever-labs-products__wever_result-validation-check",
"mcp__wever-labs-products__wever_double-charge-check",
"mcp__wever-labs-products__wever_refund-path-check",
"mcp__wever-labs-products__wever_delegated-identity-check",
"mcp__wever-labs-products__wever_purchase-record-check",
"mcp__wever-labs-products__wever_limit-substrate-check",
"mcp__wever-labs-products__wever_limit-scope-check",
"mcp__wever-labs-products__wever_limit-change-check",
"mcp__wever-labs-products__wever_threshold-review-check",
"mcp__wever-labs-products__wever_breach-path-check",
"mcp__wever-labs-products__wever_shared-cap-check",
"mcp__wever-labs-products__wever_limit-visibility-check",
"mcp__wever-labs-products__wever_limit-evidence-check",
"mcp__wever-labs-products__wever_handoff-record-check",
"mcp__wever-labs-products__wever_handoff-acceptance-check",
"mcp__wever-labs-products__wever_context-transfer-check",
"mcp__wever-labs-products__wever_receiver-identity-check",
"mcp__wever-labs-products__wever_handoff-failure-check",
"mcp__wever-labs-products__wever_chain-of-custody-check",
"mcp__wever-labs-products__wever_fleet-roster-check",
"mcp__wever-labs-products__wever_handoff-scope-check",
"mcp__wever-labs-products__wever_resume-point-check",
"mcp__wever-labs-products__wever_retry-plan-check",
"mcp__wever-labs-products__wever_partial-state-check",
"mcp__wever-labs-products__wever_rollback-path-check",
"mcp__wever-labs-products__wever_failure-notice-check",
"mcp__wever-labs-products__wever_state-reconciliation-check",
"mcp__wever-labs-products__wever_recovery-record-check",
"mcp__wever-labs-products__wever_duplicate-effect-check",
"mcp__wever-labs-products__wever_memory-scope-check",
"mcp__wever-labs-products__wever_memory-provenance-check",
"mcp__wever-labs-products__wever_memory-consent-check",
"mcp__wever-labs-products__wever_memory-deletion-check",
"mcp__wever-labs-products__wever_memory-conflict-check",
"mcp__wever-labs-products__wever_memory-sharing-check",
"mcp__wever-labs-products__wever_memory-export-check",
"mcp__wever-labs-products__wever_memory-retention-check",
"mcp__wever-labs-products__wever_delegation-scope-check",
"mcp__wever-labs-products__wever_delegation-chain-check",
"mcp__wever-labs-products__wever_delegate-vetting-check",
"mcp__wever-labs-products__wever_delegation-revocation-check",
"mcp__wever-labs-products__wever_delegation-result-check",
"mcp__wever-labs-products__wever_delegation-record-check",
"mcp__wever-labs-products__wever_delegation-expiry-check",
"mcp__wever-labs-products__wever_delegation-reporting-check",
"mcp__wever-labs-products__wever_evaluation-design-check",
"mcp__wever-labs-products__wever_evaluation-dataset-check",
"mcp__wever-labs-products__wever_evaluation-scoring-check",
"mcp__wever-labs-products__wever_evaluation-result-check",
"mcp__wever-labs-products__wever_evaluation-reproducibility-check",
"mcp__wever-labs-products__wever_evaluation-drift-check",
"mcp__wever-labs-products__wever_evaluation-cost-check",
"mcp__wever-labs-products__wever_evaluation-report-check",
"mcp__wever-labs-products__wever_negotiation-terms-check",
"mcp__wever-labs-products__wever_negotiation-mandate-check",
"mcp__wever-labs-products__wever_negotiation-counteroffer-check",
"mcp__wever-labs-products__wever_negotiation-record-check",
"mcp__wever-labs-products__wever_negotiation-fairness-check",
"mcp__wever-labs-products__wever_negotiation-deadlock-check",
"mcp__wever-labs-products__wever_negotiation-settlement-check",
"mcp__wever-labs-products__wever_negotiation-risk-check",
"mcp__wever-labs-products__wever_reputation-record-check",
"mcp__wever-labs-products__wever_reputation-source-check",
"mcp__wever-labs-products__wever_reputation-claim-check",
"mcp__wever-labs-products__wever_reputation-dispute-check",
"mcp__wever-labs-products__wever_reputation-decay-check",
"mcp__wever-labs-products__wever_reputation-portability-check",
"mcp__wever-labs-products__wever_reputation-gaming-check",
"mcp__wever-labs-products__wever_reputation-report-check",
"mcp__wever-labs-products__wever_observability-trace-check",
"mcp__wever-labs-products__wever_observability-coverage-check",
"mcp__wever-labs-products__wever_observability-alert-check",
"mcp__wever-labs-products__wever_observability-metric-check",
"mcp__wever-labs-products__wever_observability-retention-check",
"mcp__wever-labs-products__wever_observability-access-check",
"mcp__wever-labs-products__wever_observability-report-check",
"mcp__wever-labs-products__wever_knowledge-source-check",
"mcp__wever-labs-products__wever_knowledge-freshness-check",
"mcp__wever-labs-products__wever_knowledge-grounding-check",
"mcp__wever-labs-products__wever_knowledge-conflict-check",
"mcp__wever-labs-products__wever_knowledge-coverage-check",
"mcp__wever-labs-products__wever_knowledge-entry-check",
"mcp__wever-labs-products__wever_knowledge-update-check",
"mcp__wever-labs-products__wever_knowledge-report-check",
"mcp__wever-labs-products__wever_coordination-plan-check",
"mcp__wever-labs-products__wever_coordination-boundary-check",
"mcp__wever-labs-products__wever_coordination-state-check",
"mcp__wever-labs-products__wever_coordination-conflict-check",
"mcp__wever-labs-products__wever_coordination-progress-check",
"mcp__wever-labs-products__wever_coordination-roster-check",
"mcp__wever-labs-products__wever_coordination-escalation-check",
"mcp__wever-labs-products__wever_coordination-report-check",
"mcp__wever-labs-products__wever_migration-scope-check",
"mcp__wever-labs-products__wever_migration-export-check",
"mcp__wever-labs-products__wever_migration-mapping-check",
"mcp__wever-labs-products__wever_migration-loss-check",
"mcp__wever-labs-products__wever_migration-cutover-check",
"mcp__wever-labs-products__wever_migration-return-check",
"mcp__wever-labs-products__wever_migration-verify-check",
"mcp__wever-labs-products__wever_migration-report-check",
"mcp__wever-labs-products__wever_feedback-request-check",
"mcp__wever-labs-products__wever_feedback-actionability-check",
"mcp__wever-labs-products__wever_feedback-provenance-check",
"mcp__wever-labs-products__wever_feedback-incorporation-check",
"mcp__wever-labs-products__wever_feedback-closure-check",
"mcp__wever-labs-products__wever_feedback-freshness-check",
"mcp__wever-labs-products__wever_feedback-gaming-check",
"mcp__wever-labs-products__wever_feedback-report-check",
"mcp__wever-labs-products__wever_uncertainty-statement-check",
"mcp__wever-labs-products__wever_uncertainty-calibration-check",
"mcp__wever-labs-products__wever_uncertainty-evidence-gap-check",
"mcp__wever-labs-products__wever_uncertainty-assumption-check",
"mcp__wever-labs-products__wever_uncertainty-abstention-check",
"mcp__wever-labs-products__wever_uncertainty-freshness-check",
"mcp__wever-labs-products__wever_uncertainty-range-check",
"mcp__wever-labs-products__wever_uncertainty-report-check",
"mcp__wever-labs-products__wever_commitment-capture-check",
"mcp__wever-labs-products__wever_commitment-specificity-check",
"mcp__wever-labs-products__wever_commitment-tracking-check",
"mcp__wever-labs-products__wever_commitment-deadline-check",
"mcp__wever-labs-products__wever_commitment-slippage-check",
"mcp__wever-labs-products__wever_commitment-renegotiation-check",
"mcp__wever-labs-products__wever_commitment-closure-check",
"mcp__wever-labs-products__wever_commitment-report-check",
"mcp__wever-labs-products__wever_learning-capture-check",
"mcp__wever-labs-products__wever_learning-distillation-check",
"mcp__wever-labs-products__wever_learning-validation-check",
"mcp__wever-labs-products__wever_learning-application-check",
"mcp__wever-labs-products__wever_learning-retention-check",
"mcp__wever-labs-products__wever_learning-transfer-check",
"mcp__wever-labs-products__wever_learning-regression-check",
"mcp__wever-labs-products__wever_learning-report-check",
"mcp__wever-labs-products__wever_triage-intake-check",
"mcp__wever-labs-products__wever_triage-severity-check",
"mcp__wever-labs-products__wever_triage-priority-check",
"mcp__wever-labs-products__wever_triage-routing-check",
"mcp__wever-labs-products__wever_triage-aging-check",
"mcp__wever-labs-products__wever_triage-escalation-check",
"mcp__wever-labs-products__wever_triage-disposition-check",
"mcp__wever-labs-products__wever_triage-report-check",
"mcp__wever-labs-products__wever_goal-statement-check",
"mcp__wever-labs-products__wever_goal-alignment-check",
"mcp__wever-labs-products__wever_goal-conflict-check",
"mcp__wever-labs-products__wever_goal-drift-check",
"mcp__wever-labs-products__wever_goal-freshness-check",
"mcp__wever-labs-products__wever_goal-scope-check",
"mcp__wever-labs-products__wever_goal-progress-check",
"mcp__wever-labs-products__wever_goal-report-check",
"mcp__wever-labs-products__wever_vendor-identity-check",
"mcp__wever-labs-products__wever_vendor-quote-check",
"mcp__wever-labs-products__wever_vendor-terms-match-check",
"mcp__wever-labs-products__wever_vendor-price-consistency-check",
"mcp__wever-labs-products__wever_vendor-delivery-check",
"mcp__wever-labs-products__wever_vendor-refund-check",
"mcp__wever-labs-products__wever_vendor-lockin-check",
"mcp__wever-labs-products__wever_vendor-record-check",
"mcp__wever-labs-products__wever_consent-scope-check",
"mcp__wever-labs-products__wever_consent-expiry-check",
"mcp__wever-labs-products__wever_consent-revocation-check",
"mcp__wever-labs-products__wever_consent-evidence-check",
"mcp__wever-labs-products__wever_consent-purpose-check",
"mcp__wever-labs-products__wever_consent-transfer-check",
"mcp__wever-labs-products__wever_consent-renewal-check",
"mcp__wever-labs-products__wever_consent-record-check",
"mcp__wever-labs-products__wever_risk-exposure-check",
"mcp__wever-labs-products__wever_risk-reversibility-check",
"mcp__wever-labs-products__wever_risk-blast-radius-check",
"mcp__wever-labs-products__wever_risk-downside-check",
"mcp__wever-labs-products__wever_risk-concentration-check",
"mcp__wever-labs-products__wever_risk-dependency-check",
"mcp__wever-labs-products__wever_risk-timing-check",
"mcp__wever-labs-products__wever_risk-compounding-check",
"mcp__wever-labs-products__wever_onboarding-identity-check",
"mcp__wever-labs-products__wever_onboarding-credential-check",
"mcp__wever-labs-products__wever_onboarding-wallet-check",
"mcp__wever-labs-products__wever_onboarding-permissions-check",
"mcp__wever-labs-products__wever_onboarding-limits-check",
"mcp__wever-labs-products__wever_onboarding-access-check",
"mcp__wever-labs-products__wever_onboarding-readiness-check",
"mcp__wever-labs-products__wever_onboarding-record-check",
"mcp__wever-labs-products__wever_energy-tariff-check",
"mcp__wever-labs-products__wever_energy-solar-quote-check",
"mcp__wever-labs-products__wever_energy-contract-check",
"mcp__wever-labs-products__wever_energy-ev-charging-check",
"mcp__wever-labs-products__wever_energy-carbon-claim-check",
"mcp__wever-labs-products__wever_energy-bill-check",
"mcp__wever-labs-products__wever_energy-demand-charge-check",
"mcp__wever-labs-products__wever_energy-efficiency-claim-check",
"mcp__wever-labs-products__wever_dataset-quality-check",
"mcp__wever-labs-products__wever_dataset-schema-drift-check",
"mcp__wever-labs-products__wever_dataset-license-check",
"mcp__wever-labs-products__wever_dataset-synthetic-disclosure-check",
"mcp__wever-labs-products__wever_dataset-freshness-check",
"mcp__wever-labs-products__wever_dataset-lineage-check",
"mcp__wever-labs-products__wever_dataset-privacy-check",
"mcp__wever-labs-products__wever_dataset-coverage-check",
"mcp__wever-labs-products__wever_tokenized-quote-check",
"mcp__wever-labs-products__wever_tokenized-platform-terms-check",
"mcp__wever-labs-products__wever_tokenized-custody-check",
"mcp__wever-labs-products__wever_tokenized-redemption-check",
"mcp__wever-labs-products__wever_tokenized-transferability-check",
"mcp__wever-labs-products__wever_tokenized-settlement-check",
"mcp__wever-labs-products__wever_tokenized-backing-check",
"mcp__wever-labs-products__wever_tokenized-cost-check",
"mcp__wever-labs-products__wever_role-definition-check",
"mcp__wever-labs-products__wever_role-owner-check",
"mcp__wever-labs-products__wever_role-permission-match-check",
"mcp__wever-labs-products__wever_role-scope-boundary-check",
"mcp__wever-labs-products__wever_role-handoff-check",
"mcp__wever-labs-products__wever_role-review-check",
"mcp__wever-labs-products__wever_role-separation-check",
"mcp__wever-labs-products__wever_role-delegation-check",
"mcp__wever-labs-products__wever_capacity-workload-check",
"mcp__wever-labs-products__wever_capacity-queue-check",
"mcp__wever-labs-products__wever_capacity-concurrency-check",
"mcp__wever-labs-products__wever_capacity-burn-check",
"mcp__wever-labs-products__wever_capacity-backlog-check",
"mcp__wever-labs-products__wever_capacity-headroom-check",
"mcp__wever-labs-products__wever_capacity-scaling-check",
"mcp__wever-labs-products__wever_capacity-throttle-check",
"mcp__wever-labs-products__wever_travel-authorization-check",
"mcp__wever-labs-products__wever_travel-itinerary-check",
"mcp__wever-labs-products__wever_travel-passenger-check",
"mcp__wever-labs-products__wever_travel-fare-check",
"mcp__wever-labs-products__wever_travel-change-check",
"mcp__wever-labs-products__wever_travel-cancellation-check",
"mcp__wever-labs-products__wever_travel-ancillary-check",
"mcp__wever-labs-products__wever_travel-disruption-check",
"mcp__wever-labs-products__wever_insurance-coverage-check",
"mcp__wever-labs-products__wever_insurance-exclusions-check",
"mcp__wever-labs-products__wever_insurance-limits-check",
"mcp__wever-labs-products__wever_insurance-deductible-check",
"mcp__wever-labs-products__wever_insurance-claims-deadline-check",
"mcp__wever-labs-products__wever_insurance-certificate-check",
"mcp__wever-labs-products__wever_insurance-notice-check",
"mcp__wever-labs-products__wever_insurance-renewal-check"
],
"deny": [
"mcp__wever-labs-products__wever_delegated-authority",
"mcp__wever-labs-products__wever_receipt-passport",
"mcp__wever-labs-products__wever_return-package-viewer",
"mcp__wever-labs-products__wever_callback-health-monitor",
"mcp__wever-labs-products__wever_agent-sla-rail",
"mcp__wever-labs-products__wever_agent-trust-scorecard",
"mcp__wever-labs-products__wever_ap2-mandate-gateway",
"mcp__wever-labs-products__wever_agent-to-agent-handoff-pack",
"mcp__wever-labs-products__wever_proof-inbox",
"mcp__wever-labs-products__wever_rail-playground",
"mcp__wever-labs-products__wever_agent-escrow-lite",
"mcp__wever-labs-products__wever_agent-invoice-rail",
"mcp__wever-labs-products__wever_agent-permission-ledger",
"mcp__wever-labs-products__wever_agent-decision-packet-rail",
"mcp__wever-labs-products__wever_agent-file-desk-rail",
"mcp__wever-labs-products__wever_referral-rail",
"mcp__wever-labs-products__wever_subscription-rail",
"mcp__wever-labs-products__wever_workorder-create",
"mcp__wever-labs-products__wever_workorder-fund",
"mcp__wever-labs-products__wever_workorder-accept",
"mcp__wever-labs-products__wever_workorder-deliver",
"mcp__wever-labs-products__wever_workorder-release",
"mcp__wever-labs-products__wever_workorder-dispute",
"mcp__wever-labs-products__wever_payout-create",
"mcp__wever-labs-products__wever_payout-authorize",
"mcp__wever-labs-products__wever_payout-execute",
"mcp__wever-labs-products__wever_dispute-open",
"mcp__wever-labs-products__wever_dispute-submit-evidence",
"mcp__wever-labs-products__wever_dispute-resolve",
"mcp__wever-labs-products__wever_purchase-create-intent",
"mcp__wever-labs-products__wever_purchase-execute",
"mcp__wever-labs-products__wever_purchase-dispute",
"mcp__wever-labs-products__wever_sports-book-leg",
"mcp__wever-labs-products__wever_sports-cancel-leg",
"mcp__wever-labs-products__wever_accept-job",
"mcp__wever-labs-products__wever_verify-and-release",
"mcp__wever-labs-products__wever_mandate-halt",
"mcp__wever-labs-products__wever_mandate-revoke",
"mcp__wever-labs-products__wever_watch-cancel",
"mcp__wever-labs-products__wever_schedule-cancel",
"mcp__wever-labs-products__wever_digest-cancel",
"mcp__wever-labs-products__wever_kill-switch-check"
]
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as wever-labs-products. Review it before you use it.
Capability tags behind 135 rules
- mcp__wever-labs-products__wever_delegated-authority untrusted-input
- mcp__wever-labs-products__wever_agent-paid-workflow untrusted-input
- mcp__wever-labs-products__wever_unified-agent-checkout untrusted-input
- mcp__wever-labs-products__wever_diligenceops-paid-run untrusted-input
- mcp__wever-labs-products__wever_receipt-passport untrusted-input
- mcp__wever-labs-products__wever_return-package-viewer untrusted-input
- mcp__wever-labs-products__wever_agent-allowance-console untrusted-input
- mcp__wever-labs-products__wever_callback-health-monitor untrusted-input
- mcp__wever-labs-products__wever_agent-work-order-exchange untrusted-input
- mcp__wever-labs-products__wever_agent-budget-guard untrusted-input, outbound
- mcp__wever-labs-products__wever_agent-sla-rail untrusted-input
- mcp__wever-labs-products__wever_agent-trust-scorecard untrusted-input
- mcp__wever-labs-products__wever_payment-authority-inspector untrusted-input
- mcp__wever-labs-products__wever_x402-payment-gateway untrusted-input
- mcp__wever-labs-products__wever_ap2-mandate-gateway untrusted-input
- mcp__wever-labs-products__wever_agent-to-agent-handoff-pack untrusted-input
- mcp__wever-labs-products__wever_proof-inbox untrusted-input
- mcp__wever-labs-products__wever_rail-playground untrusted-input
- mcp__wever-labs-products__wever_agent-escrow-lite untrusted-input
- mcp__wever-labs-products__wever_agent-invoice-rail untrusted-input
- mcp__wever-labs-products__wever_agent-permission-ledger untrusted-input
- mcp__wever-labs-products__wever_agent-work-order-rail untrusted-input
- mcp__wever-labs-products__wever_agent-decision-packet-rail untrusted-input
- mcp__wever-labs-products__wever_agent-file-desk-rail untrusted-input
- mcp__wever-labs-products__wever_mcp-diligence-rail untrusted-input
- mcp__wever-labs-products__wever_agent-daily-operator-brief untrusted-input
- mcp__wever-labs-products__wever_proof-relay untrusted-input
- mcp__wever-labs-products__wever_referral-rail untrusted-input
- mcp__wever-labs-products__wever_subscription-rail untrusted-input
- mcp__wever-labs-products__wever_atlas-asset-passport untrusted-input
- mcp__wever-labs-products__wever_atlas-asset-search untrusted-input
- mcp__wever-labs-products__wever_treasury-balance untrusted-input
- mcp__wever-labs-products__wever_treasury-flows untrusted-input
- mcp__wever-labs-products__wever_workorder-create untrusted-input
- mcp__wever-labs-products__wever_workorder-fund untrusted-input
- mcp__wever-labs-products__wever_workorder-accept untrusted-input
- mcp__wever-labs-products__wever_workorder-deliver untrusted-input, outbound
- mcp__wever-labs-products__wever_workorder-release untrusted-input
- mcp__wever-labs-products__wever_workorder-dispute untrusted-input
- mcp__wever-labs-products__wever_payout-create untrusted-input
- mcp__wever-labs-products__wever_payout-authorize untrusted-input
- mcp__wever-labs-products__wever_payout-execute untrusted-input
- mcp__wever-labs-products__wever_payout-status untrusted-input
- mcp__wever-labs-products__wever_dispute-open untrusted-input
- mcp__wever-labs-products__wever_dispute-submit-evidence untrusted-input
- mcp__wever-labs-products__wever_dispute-resolve untrusted-input
- mcp__wever-labs-products__wever_dispute-status untrusted-input
- mcp__wever-labs-products__wever_boutique-list-products untrusted-input
- mcp__wever-labs-products__wever_boutique-confirm-delivery outbound
- mcp__wever-labs-products__wever_atlas-compare untrusted-input
- mcp__wever-labs-products__wever_atlas-rights-gap untrusted-input
- mcp__wever-labs-products__wever_boutique-real-money-acceptance untrusted-input
- mcp__wever-labs-products__wever_purchase-create-intent untrusted-input
- mcp__wever-labs-products__wever_purchase-status untrusted-input
- mcp__wever-labs-products__wever_purchase-execute untrusted-input
- mcp__wever-labs-products__wever_purchase-dispute untrusted-input
- mcp__wever-labs-products__wever_sports-plan-trip untrusted-input
- mcp__wever-labs-products__wever_sports-book-leg untrusted-input
- mcp__wever-labs-products__wever_sports-trip-status untrusted-input
- mcp__wever-labs-products__wever_sports-cancel-leg untrusted-input
- mcp__wever-labs-products__wever_matchmaking-opt-in-profile untrusted-input
- mcp__wever-labs-products__wever_matchmaking-propose-intro untrusted-input
- mcp__wever-labs-products__wever_matchmaking-approve-or-decline-intro untrusted-input
- mcp__wever-labs-products__wever_post-job untrusted-input, outbound
- mcp__wever-labs-products__wever_accept-job untrusted-input
- mcp__wever-labs-products__wever_submit-work untrusted-input
- mcp__wever-labs-products__wever_verify-and-release untrusted-input
- mcp__wever-labs-products__edge-alert untrusted-input
- mcp__wever-labs-products__fair-value untrusted-input
- mcp__wever-labs-products__wever_second-date-market-open untrusted-input
- mcp__wever-labs-products__wever_second-date-bet untrusted-input
- mcp__wever-labs-products__wever_second-date-resolve untrusted-input
- mcp__wever-labs-products__wever_mandate-issue untrusted-input
- mcp__wever-labs-products__wever_mandate-status untrusted-input
- mcp__wever-labs-products__wever_mandate-amend untrusted-input
- mcp__wever-labs-products__wever_mandate-halt untrusted-input
- mcp__wever-labs-products__wever_mandate-revoke untrusted-input
- mcp__wever-labs-products__wever_matchmaking-register-mailbox untrusted-input
- mcp__wever-labs-products__wever_data-evidence-check untrusted-input, outbound
- mcp__wever-labs-products__wever_watch-create untrusted-input, outbound
- mcp__wever-labs-products__wever_terms-evidence-check untrusted-input
- mcp__wever-labs-products__wever_counterparty-brief untrusted-input
- mcp__wever-labs-products__wever_wallet-activity-brief untrusted-input
- mcp__wever-labs-products__wever_fair-value-brief untrusted-input
- mcp__wever-labs-products__wever_listing-check untrusted-input
- mcp__wever-labs-products__wever_market-data-check untrusted-input
- mcp__wever-labs-products__wever_thesis-compare untrusted-input
- mcp__wever-labs-products__wever_contract-diff-check untrusted-input
- mcp__wever-labs-products__wever_schedule-create untrusted-input, outbound
- mcp__wever-labs-products__wever_digest-create untrusted-input, outbound
- mcp__wever-labs-products__wever_alert-route untrusted-input, outbound
- mcp__wever-labs-products__wever_changelog-check untrusted-input, outbound
- mcp__wever-labs-products__wever_pricing-evidence-check untrusted-input, outbound
- mcp__wever-labs-products__wever_status-evidence-check untrusted-input, outbound
- mcp__wever-labs-products__wever_docs-freshness-check untrusted-input, outbound
- mcp__wever-labs-products__wever_api-change-check outbound
- mcp__wever-labs-products__wever_atlas-issuer-brief untrusted-input
- mcp__wever-labs-products__wever_atlas-rights-history untrusted-input
- mcp__wever-labs-products__wever_atlas-redemption-check untrusted-input
- mcp__wever-labs-products__wever_atlas-nav-evidence untrusted-input
- mcp__wever-labs-products__wever_atlas-stage-brief untrusted-input
- mcp__wever-labs-products__wever_atlas-market-brief untrusted-input
- mcp__wever-labs-products__wever_endpoint-reliability-check untrusted-input, outbound
- mcp__wever-labs-products__wever_directory-status private-data, untrusted-input, outbound
- mcp__wever-labs-products__wever_wallet-statement untrusted-input
- mcp__wever-labs-products__wever_endpoint-honesty-check untrusted-input, outbound
- mcp__wever-labs-products__wever_payment-instruction-check outbound
- mcp__wever-labs-products__wever_invoice-payment-match outbound
- mcp__wever-labs-products__wever_payout-readiness-check outbound
- mcp__wever-labs-products__wever_payment-recovery-check outbound
- mcp__wever-labs-products__wever_escape-path-check outbound
- mcp__wever-labs-products__wever_budget-controls-check outbound
- mcp__wever-labs-products__wever_tool-exposure-check outbound
- mcp__wever-labs-products__wever_read-only-scope-check private-data, untrusted-input
- mcp__wever-labs-products__wever_external-listing-check outbound
- mcp__wever-labs-products__wever_provider-readiness-check outbound
- mcp__wever-labs-products__wever_rehearsal-check outbound
- mcp__wever-labs-products__wever_negotiation-mandate-check outbound
- mcp__wever-labs-products__wever_reputation-report-check outbound
- mcp__wever-labs-products__wever_observability-log-check private-data, untrusted-input
- mcp__wever-labs-products__wever_observability-metric-check outbound
- mcp__wever-labs-products__wever_observability-report-check outbound
- mcp__wever-labs-products__wever_coordination-escalation-check outbound
- mcp__wever-labs-products__wever_feedback-request-check outbound
- mcp__wever-labs-products__wever_vendor-identity-check outbound
- mcp__wever-labs-products__wever_consent-scope-check outbound
- mcp__wever-labs-products__wever_consent-revocation-check outbound
- mcp__wever-labs-products__wever_consent-transfer-check outbound
- mcp__wever-labs-products__wever_consent-record-check outbound
- mcp__wever-labs-products__wever_onboarding-limits-check outbound
- mcp__wever-labs-products__wever_onboarding-readiness-check outbound
- mcp__wever-labs-products__wever_role-scope-boundary-check outbound
- mcp__wever-labs-products__wever_role-handoff-check outbound
- mcp__wever-labs-products__wever_travel-passenger-check outbound
- mcp__wever-labs-products__wever_insurance-notice-check outbound
Pin
Keep the decision: fail CI when the tools change, follow every change, or show a badge.
Pin in CI
Fail a build when the tools you reviewed are not the tools the server serves today.
curl -s "https://radar.anlyon.com/api/v1/servers/io.github.CodeWever/wever-labs-products/verify?hash=8338823ade68aa42d9f6dbc01aa9f9e2f586f1ded0e2e4edd581e93ac47b010c" | jq -e '.changed == false'Pinned hash 8338823ade68, from Nov 14, 2026, 23:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.CodeWever/wever-labs-products/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.CodeWever/wever-labs-products)Source: MCP Registry · repository. Synced Nov 21, 2026, 20:17 UTC.
Wever Labs Agent Products MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the Wever Labs Agent Products MCP server safe?
- No scan can promise that. What its definitions show: 498 tools, 475 of which can change, send or delete something, and 355 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the Wever Labs Agent Products MCP server delete data?
- Yes. 42 tools can delete, cancel or revoke something: wever_delegated-authority, wever_receipt-passport, wever_return-package-viewer, wever_callback-health-monitor, wever_agent-sla-rail, wever_agent-trust-scorecard and 36 more.
Can the Wever Labs Agent Products MCP server do things that cannot be undone?
- Yes. 47 tools can send, pay, publish or merge: wever_payment-authority-inspector, wever_x402-payment-gateway, wever_policy-fit-check, wever_refund-eligibility-check, wever_payment-policy-check, wever_service-terms-check and 41 more.
What tools does the Wever Labs Agent Products MCP server have?
- 498 tools: 42 destructive, 47 irreversible, 386 write, 23 read. wever_delegated-authority, wever_receipt-passport, wever_return-package-viewer, wever_callback-health-monitor, wever_agent-sla-rail, wever_agent-trust-scorecard, wever_ap2-mandate-gateway, wever_agent-to-agent-handoff-pack and 490 more.
Does the Wever Labs Agent Products MCP server need sign-in or OAuth?
- Its remote endpoint did not answer the last anonymous look.
Does the Wever Labs Agent Products MCP server send data to outside services?
- 44 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): wever_agent-budget-guard, wever_workorder-deliver, wever_boutique-confirm-delivery, wever_post-job, wever_data-evidence-check, wever_watch-create and 38 more.
Has the Wever Labs Agent Products MCP server changed its tools?
- Yes, 10 times since Anlyon Radar started tracking it, most recently on Nov 14, 2026. Every change is on this page and in its Atom feed.