graviti MCP server
ai.thesingulariti/gravitithesingulariti.aistreamable-httpv0.2.0Scanned
Verified brand claims with receipts for agent commerce. Ranking is never paid; the engine is open.
7 tools: 3 change things and 4 only read.
- Reads private data · 1
- Reads text others write · 0
- Can send data out · 0
- No findings
Tools listed without sign-in. Remote posture
With Anlyon
3 tools here change things. Govern them.
- 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 graviti MCP server can do
The graviti MCP server is published by thesingulariti.ai, a domain the MCP Registry verified. It has 7 tools. It can create or change data (3 tools: match_intent, category_landscape and report_conversion), but cannot delete data. 4 tools only read.
1 tool reads private data (get_ledger).
None of the fixed checks matched anything in its tool definitions. That is not a safety guarantee.
Its remote endpoint lists its tools without sign-in.
Last scanned Nov 10, 2026, version 0.2.0.
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.
7 tools
Changes things 3Claude Code: ask
category_landscapeComparison-table view of every brand in a category across disclosed attributes, with verification status marked per entry.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 Comparison-table view of every brand in a category across disclosed attributes, with verification status marked per entry. Entries with captured ingredient/spec data also carry info_quality (0–100 disclosure completeness — transparency, never a rank factor; absent means not yet assessed, not zero). Every row carries evidence_state (audited / sourced / catalog_only); catalog_only rows are landscape presence only — never ranked or recommended. Rows include site_url and attributed_url (utm_source=graviti + signed gvt token — surface-level referral attribution, no user data); prefer attributed_url when linking. Rows with a published, receipt-backed discount code carry an offers array — disclosure only, never a ranking input.
Annotations the server declares
- title: Category comparison table
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
match_intentTakes a natural-language buyer question plus optional intent profile; returns recommendations ranked purely by fit score.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 Takes a natural-language buyer question plus optional intent profile; returns recommendations ranked purely by fit score. Verification status is disclosed but never affects rank. Questions outside the verified index return an explicit out-of-scope result (in_index: false), never a guess. Brands without captured evidence (catalog_only) are returned in in_landscape_not_evidenced — listed, never scored, never recommended. Each recommendation carries the brand's site link twice: site_url (raw) and attributed_url (tagged utm_source=graviti + a signed gvt token, so the brand can verify the referral — surface-level only, no user data). Prefer attributed_url when linking. Where the brand publishes a discount code/promotion on its own pages, the entry carries an offers block (code, terms, verbatim-quote receipt, 14-day re-check window) — disclosure only: offers NEVER affect ranking, and a better deal never outranks a better fit.
Annotations the server declares
- title: Match buyer intent to brands
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
report_conversionAppends one attributed-transaction event to a log — the demo stub of the pay-on-real-acquisition loop.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.
Description, as the server wrote it Appends one attributed-transaction event to a log — the demo stub of the pay-on-real-acquisition loop. It moves no money, calls no external service, and never deletes or overwrites anything; the response states where (and whether) the event was persisted on this transport. Payment only ever flows on real transactions, never on impressions or rank. Optionally pass the gvt attribution token from the recommendation link that led to the purchase: it is verified against the published Ed25519 key and recorded with the event as verifiable provenance of the originating recommendation. Optionally pass offer_code if the buyer used a published code from the index's offers block, so deal-driven conversions are attributable — recorded as reported, never a rank input.
Annotations the server declares
- title: Report an attributed conversion (demo stub)
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
Only reads 4Show 4 read-only tools
get_accountability_logPublic, machine-readable log of every seal lifecycle change: flags, degradations, revocations, and restorations, with dates and reasons — plus confirmation-loop events (event: conf…No findings
Why it is in this group
- The name "get_accountability_log" says "get": a word for reading.
Description, as the server wrote it Public, machine-readable log of every seal lifecycle change: flags, degradations, revocations, and restorations, with dates and reasons — plus confirmation-loop events (event: confirmation_request / confirmation_resolved), where the audit asked a brand to confirm a claim it couldn't re-verify: a visible ask, never a penalty. Also logged: claim_removed (a claim struck from a record, reason public), gap_report_delivered (a paid-tier report delivered to a brand, payment status disclosed — findings and rank never affected), and roster events (brand_added with lane/source, brand_removed, brand_declined, brand_skipped — every future change to who is in the index lands here per brand, reason public). The seal can be lost — a pulled seal is the loudest proof it means something. Payment can fast-track a re-evaluation; it can never change what it finds.
Annotations the server declares
- title: Public accountability log
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_gap_reportWhat stands between a brand and top-of-category: attribute gaps vs the category leader, open flags, cohort blind spots per buyer-intent voice, and the ingredient/spec disclosure di…No findings
Why it is in this group
- The name "get_gap_report" says "get": a word for reading.
Description, as the server wrote it What stands between a brand and top-of-category: attribute gaps vs the category leader, open flags, cohort blind spots per buyer-intent voice, and the ingredient/spec disclosure dimension — the brand's info_quality score with the exact publishable fixes that would raise it (publishing more truth, not paying Graviti). Demo of the paid brand-intelligence product — computed deterministically from the index; in production also draws on consented, compensated cohort signals (profiles never leave the user's agent). Catalog-only brands (no captured claims) receive an insufficient-evidence report instead of ranks.
Annotations the server declares
- title: Brand gap report (brand intelligence)
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_ledgerAppend-only version history of the index: each entry commits to the SHA-256 of the canonical index content and the previous entry's hash (rewriting history breaks every hash after …Reads private dataNo findings
Why it is in this group
- The name "get_ledger" says "get": a word for reading.
What it can reach
- private-dataThe description's first sentence says "commit": it reads repositories.
Description, as the server wrote it Append-only version history of the index: each entry commits to the SHA-256 of the canonical index content and the previous entry's hash (rewriting history breaks every hash after it), is Ed25519-signed, and is anchored to Bitcoin via OpenTimestamps. Returns entries plus independent verification instructions. Even Graviti cannot rewrite the record.
Annotations the server declares
- title: Integrity ledger (hash-chained, signed, Bitcoin-anchored)
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_verified_claimsStructured claims for one brand: provenance URLs, freshness timestamps, corroboration counts, and verification status (verified means the brand pays for auditing — always disclosed…No findings
Why it is in this group
- The name "get_verified_claims" says "get": a word for reading.
Description, as the server wrote it Structured claims for one brand: provenance URLs, freshness timestamps, corroboration counts, and verification status (verified means the brand pays for auditing — always disclosed). Where captured, includes the specifications block (ingredient/spec disclosure with per-fact provenance) and its info_quality score — a 0–100 measure of disclosure completeness that never influences ranking. Graviti does not assess efficacy or clinical outcomes (not_assessed fence in every block). Claims with confirmation_status 'awaiting_brand_confirmation' are pending the brand's response to a re-verification ask — treat them as unconfirmed, not as violations. Where captured, an offers block lists the brand's published discount codes with receipts and full status history (active / expired / failed_verification) — disclosure only, never a ranking input.
Annotations the server declares
- title: Get claims with provenance for a brand
- 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 or outbound.
- 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 out0
- 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, 18:28 UTC, unchanged since Oct 22, 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
- yes
- Tools that change things, listed without credentials
- 3
- Protocol version
- 2025-06-18
- 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
- WarnThe whole server
The endpoint served tools/list without credentials, and the list has 3 tools that change things (3 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://graviti.thesingulariti.ai/api/mcp: tools/list returned 7 tools to an anonymous client, including match_intent (write), category_landscape (write), report_conversion (write)
Findings
Matches of fixed checks, with the text they matched.
No check matched anything in these tool definitions. That is not a safety guarantee.
History
Changes to its tool definitions and to its remote endpoint, newest first.
No change seen yet. Its tools are rescanned every day, and a remote endpoint is looked at once a week.
Install
Pinned to version 0.2.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http graviti https://graviti.thesingulariti.ai/api/mcpclaude mcp add --transport http graviti https://graviti.thesingulariti.ai/api/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__graviti__get_verified_claims",
"mcp__graviti__get_accountability_log",
"mcp__graviti__get_ledger",
"mcp__graviti__get_gap_report"
],
"ask": [
"mcp__graviti__match_intent",
"mcp__graviti__category_landscape",
"mcp__graviti__report_conversion"
],
"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 graviti. Review it before you use it.
Capability tags behind 1 rule
- mcp__graviti__get_ledger 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/ai.thesingulariti/graviti/verify?hash=d83058454df6a29e1d243f746987c5937091637e3fcae2841d269437ee93271a" | jq -e '.changed == false'Pinned hash d83058454df6, from Nov 10, 2026, 04:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/ai.thesingulariti/graviti/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/ai.thesingulariti/graviti)Source: MCP Registry · repository. Synced Oct 21, 2026, 17:43 UTC.
graviti MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the graviti MCP server safe?
- No scan can promise that. What its definitions show: 7 tools, 3 of which can change, send or delete something, and 0 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 graviti MCP server delete data?
- No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.
Can the graviti MCP server do things that cannot be undone?
- No tool is classed as irreversible: none of them sends, pays, publishes or merges, going by their definitions.
What tools does the graviti MCP server have?
- 7 tools: 3 write, 4 read. match_intent, category_landscape, report_conversion, get_verified_claims, get_accountability_log, get_ledger and get_gap_report.
Does the graviti MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the graviti MCP server send data to outside services?
- None of its tools sends, posts or publishes, or takes a recipient, channel or URL.
Has the graviti MCP server changed its tools?
- Not since Anlyon Radar started tracking it. Last scanned Nov 10, 2026.