Skip to content

attest — verification for AI agent artifacts

by mrpkk · remote · no sign-in to list tools · v0.1.0 · published by GitHub user mrpkk

Verify an AI agent artifact: schema, emptiness, provenance and mandate authority, Ed25519-signed.

Changes · 3
CompareRegistry entry

3 tools: 3 change things.

Scanned Oct 21, 2026

No check matched anything in the tool definitions. No finding is not a safety guarantee. How the checks work.

claude mcp add --transport http attest https://attest.rta.lat/mcp

Tools

Open a tool for its reasons and full description. Descriptions are the publisher's own words, shown as they are.

3 tools

Changes things 3Claude Code: ask
  • bridge_checkSettle a job between two agents through an escrow bridge.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
    Settle a job between two agents through an escrow bridge. The client posts a task with a budget held in escrow, the worker delivers an artifact, the bridge verifies it and issues a signed RECEIPT. The budget is released only after verification passes. The RECEIPT is verifiable by a third party without access to the bridge, which is what makes it portable between worker, client and insurer. Actions: job (create), deliver (submit and verify), release (accept), refund (reject), verify (check someone else's receipt), trust (reputation). Use when an agent takes on paid work, or when you need to prove to a third party that the work was done.

    Annotations the server declares

    • title: Bridge: task to work to money
    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: true
  • verify_artifactVerify an artifact produced by an AI agent before accepting or paying 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
    Verify an artifact produced by an AI agent before accepting or paying for it. Four independent layers: schema conformance, whether there is anything inside (uniform lists, stubs, truncations), provenance bound to the artifact hash, and whether a signed mandate permitted the action. Returns a verdict (accept/review/reject), a 0-100 trust score, concrete findings and an Ed25519 signature. Rule-based, no LLM: the same artifact always yields the same result. Use before accepting work from another agent and before payment.

    Annotations the server declares

    • title: Verify an AI agent artifact
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • verify_artifact_batchVerify several artifacts at once — a scan report, a list of findings, a set of tool descriptions.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
    Verify several artifacts at once — a scan report, a list of findings, a set of tool descriptions. One call, one verdict per artifact, plus a summary of how many were accepted, need review, or were rejected. Use when an agent produced a batch (twenty findings from a security scan, for example) and each of them has to be checked before the work is paid for.

    Annotations the server declares

    • title: Verify a batch of AI agent artifacts
    • 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.

Install

Pinned to version 0.1.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http attest https://attest.rta.lat/mcp

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [],
    "ask": [
      "mcp__attest__verify_artifact",
      "mcp__attest__bridge_check",
      "mcp__attest__verify_artifact_batch"
    ],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. It assumes the server is installed as attest. Review it before you use it.