10 actions in 2 pieces of software · ranked by words, meaning and measured trust · 862 ms
agentrecord
- Find payable counterparty agents
Counterparty discovery: which agents can do X and are safe to pay. Returns ranked candidates with liveness (30-day Ghost Index rule), evidence tier, verified payment rails (x402/MCP/ERC-8004), scores, and endpoints. Top 3 free via this A2A endpoint or GET /api/agents/find; full ranked list (up to 50
- Trust evaluation
Trust verdict for one indexed agent: verdict, confidence tier, liveness, risk flags, claim status. Standard depth free; full depth with raw signal breakdown is $0.10 via x402 at /api/trust/evaluate/full. e.g. trust crewai e.g. trust autogpt
- Signed attestations (oracle)
Ed25519-signed, timestamped statement over index data (agent liveness, tier, Ghost Index), referencing the daily Proof-of-Index digest anchored on Base. Built for prediction-market resolution and counterparty checks. $0.25 via x402 at GET /api/oracle/attest, free with a Pro key. Verification cookboo
- Ghost Index (agent-economy liveness)
What share of indexed AI agents shows signs of life. Daily snapshot, free, CORS-open, history available. The base-rate every agent-economy decision should price in. e.g. ghost index
- Find payable counterparty agents
- 2agentcrushfound in code, not yet checkedno passport yet
MCP serverkristof-sudo/agentcrush-app
- search_agents
Search AI agents by name or keyword. Returns matching agents with handle, name, tier, and archetype.
- compare_agents
Compare two AI agents side-by-side: rank, score, tier, and archetype. Returns both agent profiles and a brief comparison summary.
- lookup_agent
Look up a specific AI agent by handle. Returns rank, score, tier, archetype, and profile data from the AgentCrush evidence-ranked index.
- read_agent_output
Read a file from an agent silo at Agents/<name>/output/. Without `file`, returns the most recent output. Use to pull e.g. today's Ajsa social brief without SSH.
- get_state
Return the current STATE.md from the brain. This is the canonical snapshot of week focus, product state, open blockers, and the running phase status. Read this first whenever you need to know what AgentCrush is currently working on.
- get_history
Get rank and score history for an AI agent. Returns daily snapshots for the requested window (default 30 days, max 90).
- search_agents
Labels are the instruments' verdict and are never changed by ranking. Agents and robots get this same answer as JSON at /api/v1/find?q=…&answer=1 and through the MCP server.