Skip to content

Mifori CertAgent Trust Oracle

by mifori · remote · no sign-in to list tools · v1.0.0 · published by GitHub user mifori

Live AI agent trust oracle: audit synthetic IQ, latency & JSON compliance via x402 or free tokens.

Changes · 2
CompareRegistry entry

4 tools: 2 change things and 2 only read.

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 certagent https://agent.mifori.com/mcp?v=1

Tools

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

4 tools

Changes things 2Claude Code: ask
  • submit_agent_feedbackPersists machine-to-machine telemetry, score disputes, or initial sandbox credit requests in <100ms with no authentication or payment required, immediately granting 1 free single-u…No findings

    Why it is in this group

    • The name "submit_agent_feedback" says "submit": a word for a change.
    Description, as the server wrote it
    Persists machine-to-machine telemetry, score disputes, or initial sandbox credit requests in <100ms with no authentication or payment required, immediately granting 1 free single-use audit credit token ('mfr_fbk_...'). First-time evaluators without a prior audit or dispute can pass eval_id='sandbox' (or 'onboarding') and describe their intended evaluation in reported_issue to receive a legitimate free reward_token for verify_agent_reputation without fabricating a complaint.

    Annotations the server declares

    • title: Submit Telemetry or Request Free Sandbox Audit Credit
    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • verify_agent_reputationExecutes a live multi-protocol audit against target_bot_url (supporting MCP JSON-RPC 2.0 initialize + tools/list + execution probe over HTTP/SSE, SEP-1649 MCP Server Cards, Google …No findings

    Why it is in this group

    • The description's first verb is "execute": a word for a change.
    Description, as the server wrote it
    Executes a live multi-protocol audit against target_bot_url (supporting MCP JSON-RPC 2.0 initialize + tools/list + execution probe over HTTP/SSE, SEP-1649 MCP Server Cards, Google A2A Agent Cards, and Mifori Cognitive challenge-response endpoints; non-agent static REST URLs are strictly rejected with 0 points) to benchmark Synthetic IQ (0-100), steady-state latency, JSON schema compliance, and verified tool capabilities, returning an Ed25519 asymmetrically signed certificate (eval_id). Pass public=false to run a private/confidential audit that returns a signed certificate without publishing target_bot_url to the public leaderboard. Two-step parameter workflow: (1) call with target_bot_url alone (omitting payment_proof) for a free PAYMENT_REQUIRED_402 quote, or call submit_agent_feedback with eval_id='sandbox' to obtain a free single-use 'mfr_fbk_...' credit token without needing a prior dispute; (2) invoke with payment_proof set to a 'mfr_fbk_...' token or a verified $0.005 Base Mainnet USDC transaction hash ('0x...').

    Annotations the server declares

    • title: Verify Agent Reputation (Multi-Protocol Audit)
    • readOnlyHint: false
    • destructiveHint: false
    • idempotentHint: false
    • openWorldHint: true
Only reads 2Claude Code: allow
  • get_agent_certificateRetrieves and cryptographically validates a signed audit certificate by eval_id (including private certificates created with public=false) or lists chronological public audit histo…No findings

    Why it is in this group

    • The name "get_agent_certificate" says "get": a word for reading.
    Description, as the server wrote it
    Retrieves and cryptographically validates a signed audit certificate by eval_id (including private certificates created with public=false) or lists chronological public audit history for an agent_id in <50ms with no authentication, payment, or rate limits. Pass eval_id alone for an exact single-certificate lookup with asymmetric Ed25519 signature verification against Mifori's public key (/.well-known/jwks.json), or pass agent_id (a DID or target_bot_url) with limit (1-50, default 10) to inspect historical evaluations.

    Annotations the server declares

    • title: Get Agent Audit Certificate & History
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • get_agent_leaderboardQueries a free, unauthenticated snapshot of verified MCP servers and autonomous AI agents deduplicated by endpoint and ordered by Synthetic IQ descending (0-100) and round-trip lat…No findings

    Why it is in this group

    • The name "get_agent_leaderboard" says "get": a word for reading.
    Description, as the server wrote it
    Queries a free, unauthenticated snapshot of verified MCP servers and autonomous AI agents deduplicated by endpoint and ordered by Synthetic IQ descending (0-100) and round-trip latency ascending (ms). Each entry includes protocol_detected ('MCP 2.0', 'Mifori Cognitive v1', etc.) and verified_capabilities (listing verified MCP tool names and server versions). Pass query (e.g. 'wiki', 'search', 'cloudflare', 'crypto', 'docs') to filter agents by tool capability, DID, or domain, and optionally filter by tier ('AAA', 'AA', 'B', 'REJECTED').

    Annotations the server declares

    • title: Search & Get Verified Agent Leaderboard
    • 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 1.0.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http certagent https://agent.mifori.com/mcp?v=1

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": [
      "mcp__certagent__get_agent_leaderboard",
      "mcp__certagent__get_agent_certificate"
    ],
    "ask": [
      "mcp__certagent__verify_agent_reputation",
      "mcp__certagent__submit_agent_feedback"
    ],
    "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 certagent. Review it before you use it.