Crest Counterparty Intelligence

Counterparty, credit, and token-risk intelligence for the x402 agent economy. Crest scores the code AND the people behind it, from a settlement graph no directory has.

Live MCP endpoint: https://data.crestsystems.ai/mcp (Streamable HTTP, no auth). On Smithery: crest/counterparty-intelligence.

Why

Before an agent pays a wallet, buys a token, or extends another agent credit, it needs to know what is on the other side. Most scanners read only the contract or only the balance. Crest reads the contract's powers (what it can do) and the reputation of the wallet that controls it (who holds the keys), and gates one by the other. The same powers are safe under a trusted issuer and a rug under an anonymous, drained wallet.

Related MCP server: AgentScore

Tools

All tools are free and read-only. The deep scored reports are $0.05 x402 calls on Base. Not financial advice.

Examples

token_risk(0x4ed4e862860bed51a9570b96d89af5e1b0efefed)
  -> { risk_level: "low", attack_surface: 20, controller_trust: 0.65 }

crest_score(0xe4a961cf1ee5b305561e38c69ed9dbd058ff036e)
  -> { crest_score: 89, band: "A", decision: "ok_to_extend" }

Run it yourself

This repository is a thin, read-only client over Crest's public preview endpoints. No keys, no database, no secrets.

npm install
node index.mjs        # listens on :8080, MCP at POST /mcp

Add to an MCP client (e.g. Claude Code) pointing at the hosted endpoint:

claude mcp add --transport http crest "https://data.crestsystems.ai/mcp"

About

By Crest Deployment Systems. The de-noised state of the x402 economy is at crestsystems.ai/x402; token risk at crestsystems.ai/token; agent credit at crestsystems.ai/credit.

License

MIT