Cred Protocol
Book a demo

Trust signals for agents and the wallets they serve.

Check a wallet before an agent pays it, lends to it or lets it in: its credit score, whether it is on a sanctions list, and how likely it is to be a sybil. Over MCP, the REST API, or pay per call with x402 and MPP, no account needed.

Book a demoGet an API key
mcp · tools/callsample
{ "name": "get_credit_score",
  "arguments": { "address": "0x7a3f…c91e" } }

→ { "score": 862, "range": "very_good",
    "model_version": "andromeda_1.3" }

Three checks before an agent acts.

Each is one call, priced in Cred Units: 1 for a score, 1 for a sanctions screen, 3 for a sybil check.

  1. Step 1

    Is it creditworthy?

    Score the wallet and gate by band: who gets access, credit or a better rate.

    get_credit_score · GET /score/address/{address}

  2. Step 2

    Is it sanctioned?

    Screen it against the OFAC lists, refreshed from the Treasury every hour.

    screen_sanctions · GET /sanctions/screen/{address}

  3. Step 3

    Is it one of many?

    A sybil analysis for airdrops, allowlists and anything that should go to one person once.

    get_sybil_score · GET /identity/address/{address}/sybil

Built for software that pays its own way.

  • No account neededPay per call in USDC over x402 on Base or SKALE, or over MPP; a 402 response tells the agent the price first.
  • Works where agents runAn MCP server for any MCP client, plus TypeScript (Express, Hono, Next.js) and Python (FastAPI) SDKs.
  • Sandbox firstScore and report endpoints have sandbox twins with mock wallets that cost no Cred Units, so an agent can be tested before it spends.

Questions

Do I need an account?

No. An agent can pay per call over x402 or MPP. With an account you get an API key, monthly Cred Units on a plan, and the developer console.

What does a call cost?

A Cred Unit is $0.01. A score or a sanctions screen is 1 CU, a sybil check 3, and a full credit report 7. Batch calls are priced per address, up to 50 per call.

Can the score gate access in my app?

Yes. Key access on the range band (low to excellent) rather than the raw number, and use the factor ratings to tell a user what would change the outcome.

See your own wallets scored.

A 30-minute call with live scoring on wallets you choose, and the factor reasons behind every number. No commitment.

Building an AI agent? Get an API key