agiscorecard-web3-workbench MCP server
io.github.f-tiger/agiscorecard-web3-workbenchf-tigerstreamable-httpv1.7.0Scanned
AGI Scorecard Web3 Workbench: ten free deterministic review tools (stablecoin, gas, zkML, agents).
12 tools: 1 do something that can't be undone, 6 change things and 5 only read.
- Reads private data · 1
- Reads text others write · 0
- Can send data out · 1
- 2 findings
Sign-in required, no OAuth metadata. Remote posture
With Anlyon
1 tool here does something that can't be undone. Govern it.
- Anlyon lists the tools and adopts none.
- Each tool you adopt gets approvals and receipts.
- Your agent never holds the key.
A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.
What the agiscorecard-web3-workbench MCP server can do
The agiscorecard-web3-workbench MCP server is published by the GitHub account f-tiger. It has 12 tools. It can do things that cannot be undone, such as sending, paying or publishing (decode_token_call) and create or change data (6 tools: reconcile_stablecoin_records, review_agent_evidence, plan_fallback_routes, check_data_permissions and 2 more), but cannot delete data. 5 tools only read.
1 tool reads private data (compare_rwa_disclosures); 1 tool can send data out of your system (decode_token_call).
The fixed checks matched 2 findings (annotation-conflict). Each one comes with the text it matched.
Last scanned Nov 21, 2026, version 1.7.0.
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.
12 tools
Can't be undone 1Claude Code: ask
decode_token_callCall Lens decodes supported token-call shapes locally and compares the recipient, spender and atomic amount with your expectations.Can send data out1 finding
Why it is in this group
- The name "decode_token_call" says "call": a word for a change.
- The input addresses people (input.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.to): what the tool writes goes there.
Findings
- Warn
The server marks "decode_token_call" 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":true,"openWorldHint":false}; The name "decode_token_call" says "call": a word for a change.
Description, as the server wrote it Call Lens decodes supported token-call shapes locally and compares the recipient, spender and atomic amount with your expectations. It distinguishes ERC-20 allowances from ERC-721 token IDs using the declared standard. It never connects a wallet or evaluates contract safety. No chain simulation, malicious-contract detection, token identity lookup, balance inspection or safety verdict. Retrieve calls with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
Changes things 6Claude Code: ask
check_data_permissionsData Permit compares an intended AI use with your recorded grant conditions: region, expiry, revocation and consent status.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 Data Permit compares an intended AI use with your recorded grant conditions: region, expiry, revocation and consent status. It calculates a proposed split of received revenue and keeps incompatible shares unallocated. Matching fields does not establish a legal license. No legal opinion, rights verification, consent collection, access enforcement, token issuance or payment. Retrieve permit with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
model_contributor_rewardsIncentive Lab separates customer revenue from token subsidy, then allocates a declared reward budget using accepted work and quality weights.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 Incentive Lab separates customer revenue from token subsidy, then allocates a declared reward budget using accepted work and quality weights. Contributor caps and exact budget conservation keep the scenario inspectable. The worksheet does not establish Sybil resistance or distribute tokens. No live subnet data, fraud detection, Bittensor emission replication, staking recommendation or token distribution. Retrieve incentives with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
plan_fallback_routesRoute Lab compares short supplier sequences using entered cost, success and latency assumptions.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 Route Lab compares short supplier sequences using entered cost, success and latency assumptions. It checks the worst-case total against your budget before sorting feasible options by expected cost. It is an offline planning tool; no model or payment is called. Offline planning only. No live requests, model judge, payment, dynamic bandit or production routing. Retrieve route with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
plan_zkml_costsProof Plan estimates proving compute and onchain verification costs from your own measurements.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 "plan_zkml_costs" 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":true,"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 Proof Plan estimates proving compute and onchain verification costs from your own measurements. It checks a per-job latency target and preserves model, input and verification-key hashes. It plans an integration; it does not generate or verify a zero-knowledge proof. No ZK proof generation or cryptographic verification. File hashing verifies bytes only; it is not proof of model correctness. Retrieve proof with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
reconcile_stablecoin_recordsStable Reconcile compares your declared invoices with recorded stablecoin transfers.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 Stable Reconcile compares your declared invoices with recorded stablecoin transfers. It separates partial payments, pending transfers and wrong-network records, with exact decimal totals. Use it for a local exception review before updating your books. No wallet sync, automatic matching, FX, refund processing, tax accounting or independent finality checks. Retrieve reconcile with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
review_agent_evidenceAgent Evidence reviews imported task outcomes within one task, version and date window.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 Evidence reviews imported task outcomes within one task, version and date window. It deduplicates sample labels and shows failures, exclusions and uncertainty. It helps review a supplier comparison; it does not run evaluations or verify reviewer identities. No public reputation ranking, identity verification, calibrated prediction or Sybil detection. Retrieve evidence with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
Only reads 5Show 5 read-only tools
compare_compute_costsCompute Lens compares measured workload runs using total entered compute, setup, storage and egress costs per accepted output.No findings
Why it is in this group
- The name "compare_compute_costs" says "compare": a word for reading.
Description, as the server wrote it Compute Lens compares measured workload runs using total entered compute, setup, storage and egress costs per accepted output. It filters mismatched workload labels, currencies, quality and memory limits. All examples are fictional; no live GPU quotes are supplied. No live GPU inventory, automatic benchmarking, reservation, infrastructure management or verified savings. Retrieve compute with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
compare_rwa_disclosuresRWA Notes compares two sets of supplied asset disclosure fields while retaining units, dates and source references.Reads private dataNo findings
Why it is in this group
- The name "compare_rwa_disclosures" says "compare": a word for reading.
What it can reach
- private-dataThe description's first sentence says "note": it reads notes.
Description, as the server wrote it RWA Notes compares two sets of supplied asset disclosure fields while retaining units, dates and source references. It flags missing values and incompatible comparisons. It does not verify reserves, issuer solvency or redemption rights. No live asset feed, reserve audit, redemption verification, credit rating or investment recommendation. Retrieve disclosures with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
fetchRetrieve a public tool reference by the ID returned from search.No findings
Why it is in this group
- The name "fetch" says "fetch": a word for reading.
Description, as the server wrote it Retrieve a public tool reference by the ID returned from search. Contains methodology, limits, official source links and fictional worked examples; no user records. Cite the returned canonical URL.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
read_protocol_profilesProtocol Ledger is a free, dated reference for five payment, agent-identity and token interfaces.No findings
Why it is in this group
- The name "read_protocol_profiles" says "read": a word for reading.
Description, as the server wrote it Protocol Ledger is a free, dated reference for five payment, agent-identity and token interfaces. Each record preserves its official specification, review date and interpretation limits. Filter the same snapshot in the browser or retrieve it through the JSON API. Five initial profiles; no market prices, exhaustive schema coverage, real-time protocol monitoring or paid endpoint. Retrieve protocol with fetch or read its example resource to obtain exact inputs. Parameters are processed remotely without application persistence.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
searchSearch this server’s public AI/Web3 tools and citation pages.No findings
Why it is in this group
- The name "search" says "search": a word for reading.
Description, as the server wrote it Search this server’s public AI/Web3 tools and citation pages. English and Chinese names supported. Empty query lists the available worksheets.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- 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.
No private-data, untrusted-input, outbound path in this server alone: no tool is tagged untrusted-input.
- Reads private data1
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write0
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out1
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
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, 19:27 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
- 429
- 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
- not observed
- Authorization servers
- not observed
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- not observed
- Client ID metadata documents
- not observed
The probe stopped: initialize answered HTTP 429.
Posture changes
- Nov 20, 2026
- HTTP status of an anonymous initialize: 200 to 429
- tools/list served without credentials: yes to no
- Tools that change things, listed without credentials: 7 to 0
- Protocol version: 2025-06-18 to none
Findings
Matches of fixed checks, with the text they matched.
- Warnplan_zkml_costs
The server marks "plan_zkml_costs" 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":true,"openWorldHint":false}; The input takes a statement or command to run (input.statement), which can change anything it can reach.
The server marks "decode_token_call" 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":true,"openWorldHint":false}; The name "decode_token_call" says "call": a word for a change.
History
Changes to its tool definitions and to its remote endpoint, newest first.
PostureRemote posture changed. HTTP status of an anonymous initialize: 200 -> 429; tools/list served without credentials: yes -> no; Tools that change things, listed without credentials: 7 -> 0 (and 1 more).View diff
- HTTP status of an anonymous initialize: 200 to 429
- tools/list served without credentials: yes to no
- Tools that change things, listed without credentials: 7 to 0
- Protocol version: 2025-06-18 to none
Install
Pinned to version 1.7.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http agiscorecard-web3-workbench https://web3.agiscorecard.com/mcpclaude mcp add --transport http agiscorecard-web3-workbench https://web3.agiscorecard.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__agiscorecard-web3-workbench__search",
"mcp__agiscorecard-web3-workbench__fetch",
"mcp__agiscorecard-web3-workbench__read_protocol_profiles",
"mcp__agiscorecard-web3-workbench__compare_compute_costs",
"mcp__agiscorecard-web3-workbench__compare_rwa_disclosures"
],
"ask": [
"mcp__agiscorecard-web3-workbench__reconcile_stablecoin_records",
"mcp__agiscorecard-web3-workbench__review_agent_evidence",
"mcp__agiscorecard-web3-workbench__plan_fallback_routes",
"mcp__agiscorecard-web3-workbench__check_data_permissions",
"mcp__agiscorecard-web3-workbench__model_contributor_rewards",
"mcp__agiscorecard-web3-workbench__plan_zkml_costs",
"mcp__agiscorecard-web3-workbench__decode_token_call"
],
"deny": []
}
}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 agiscorecard-web3-workbench. Review it before you use it.
Capability tags behind 2 rules
- mcp__agiscorecard-web3-workbench__decode_token_call outbound
- mcp__agiscorecard-web3-workbench__compare_rwa_disclosures private-data
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.f-tiger/agiscorecard-web3-workbench/verify?hash=9b5dba62900696a8391f970d9cdc35ea2e4d037a4a5e5fe084ea024df086ce58" | jq -e '.changed == false'Pinned hash 9b5dba629006, from Nov 21, 2026, 12:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/io.github.f-tiger/agiscorecard-web3-workbench/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/io.github.f-tiger/agiscorecard-web3-workbench)Source: MCP Registry. Synced Oct 21, 2026, 18:33 UTC.
agiscorecard-web3-workbench MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the agiscorecard-web3-workbench MCP server safe?
- No scan can promise that. What its definitions show: 12 tools, 7 of which can change, send or delete something, and 2 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 agiscorecard-web3-workbench MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the agiscorecard-web3-workbench MCP server do things that cannot be undone?
- Yes. 1 tool can send, pay, publish or merge: decode_token_call.
What tools does the agiscorecard-web3-workbench MCP server have?
- 12 tools: 1 irreversible, 6 write, 5 read. decode_token_call, reconcile_stablecoin_records, review_agent_evidence, plan_fallback_routes, check_data_permissions, model_contributor_rewards, plan_zkml_costs, search and 4 more.
Does the agiscorecard-web3-workbench MCP server need sign-in or OAuth?
- Its remote endpoint did not answer the last anonymous look.
Does the agiscorecard-web3-workbench MCP server send data to outside services?
- 1 tool can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): decode_token_call.
Has the agiscorecard-web3-workbench MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 21, 2026.