OFAC screening API integration patterns
A single HTTP call returns ALLOW or BLOCK, and it plugs in via x402, MCP, or agent SDKs.
TL;DR
TL;DR: OFAC screening is one HTTP call under 100 ms that returns ALLOW or BLOCK, and you can integrate it directly, through x402 at payment time, via an MCP server, or through agent SDKs.
The core call
The fundamental pattern is simple: send a counterparty identifier (name, wallet, or jurisdiction) and receive a decision. Clean means ALLOW, flagged means BLOCK. The check runs against live data of 947 OFAC-listed wallets, 19,218 SDN names, and 16 jurisdictions, synced hourly.
Integration patterns
- Direct HTTP: one call inside any backend or middleware.
- MCP server: pip install sanctions-mcp for agent tool loops.
- x402: $0.05 per check at payment time.
- Agent SDKs: LangChain, CrewAI, Eliza, OpenAI Agents SDK, Coinbase AgentKit.
Under the hood
Whatever the pattern, the four tools stay the same: sanctions_check for the decision, risk_score for context, kya_verify for counterparty documentation, and dispute_open for review. The 4-Gate protocol (SCREEN, SCORE, STOP, STAMP) describes the full flow any integration should implement.