AI assistants are great at reasoning about on-chain data — but they can't see whether a wallet is sanctioned, tied to a scam, or where its funds came from unless you give them a tool. PublicAML now ships an MCP server, so Claude and ChatGPT can run AML/KYT checks directly in the chat.
What it does
Once connected, the assistant can call three tools against PublicAML's keyless backend:
- screen_address — risk score, sanctions status, category and source of funds for a BTC/ETH/BNB Chain/TRON address
- trace_funds — follow where funds moved
- list_counterparties — who the address transacted with
No API key, no account — it is a non-profit public good.
Two ways to add it
Remote (zero install) — add the hosted server as a custom connector:
https://mcp.publicaml.org/mcp
In Claude: Settings → Connectors → Add custom connector → paste the URL. In ChatGPT: add it as a custom MCP connector. Then ask "screen 0x28C6…d60 for AML risk" and it calls the tool.
Local (stdio) — for developers who want it in their own MCP client:
npx @publicaml/mcp-server
Both are listed in the official MCP registry (search "publicaml" at registry.modelcontextprotocol.io).
Why it matters
Pre-transaction screening usually means a paid API or a separate dashboard. Putting it behind MCP means the check happens where you already are — mid-conversation, mid-investigation — for free.
Try the API directly: https://publicaml.org/api · Free wallet check: https://publicaml.org/free-crypto-aml/
Top comments (0)