AI infrastructure

OFAC MCP server for AI agents

Every CompliAPI screening capability is exposed as a tool over the Model Context Protocol, so AI agents and assistants can run OFAC checks directly — same API token, same quota, same request log as the REST API.

Get your free API key Read the documentation

https://api.compliapi.com/mcp

{
  "mcpServers": {
    "compliapi": {
      "url": "https://api.compliapi.com/mcp",
      "headers": { "Authorization": "Bearer capi_..." }
    }
  }
}

Eight screening tools

screen_crypto_address, screen_email, screen_website, screen_government_id, screen_country, screen_pep, detect_vpn and geolocate_ip.

Standard MCP transport

Served over streamable HTTP at api.compliapi.com/mcp — works with any MCP client that supports remote servers.

Same credentials as the REST API

Authenticate with your existing CompliAPI token. No separate agent accounts to manage.

Same metering and logging

Tool calls are metered against your organization's quota and recorded exactly like REST requests.

Nothing to host

The server runs as part of CompliAPI. Point your agent's config at the URL and go.

Screening your agents can call natively

If your workflows are increasingly agent-driven, wiring compliance checks into them usually means writing and maintaining custom tool wrappers around a REST API for every agent framework you use.

CompliAPI ships the MCP server for you. Any MCP-capable agent — coding assistants, support agents, payment-ops automations — gets the full set of screening tools by adding one entry to its configuration, and every call lands in the same organization-level quota and request log as your application traffic.

The MCP server exposes the same screening capabilities as the REST API, with the same scope: direct checks against OFAC SDN data plus geolocation and VPN signals. It does not make compliance decisions for your agents.

How it works

Add the server to your agent's config
Point any MCP client at https://api.compliapi.com/mcp with your API token as a bearer header.
The agent discovers and calls the tools
Tools like screen_crypto_address and screen_country become available to the agent alongside its other capabilities.
Results are structured and logged
Tool calls return the same structured screening results as the REST API, metered against your organization's quota and recorded in your request log.

What you can screen

Crypto addresses and ENS names (screen_crypto_address)
Email addresses (screen_email)
Websites and domains (screen_website)
Government-issued IDs (screen_government_id)
Countries and regions (screen_country)
People, against the OpenSanctions PEP dataset (screen_pep)
IP addresses (detect_vpn, geolocate_ip)

Where teams use it

Agent-driven onboarding

An agent processing new accounts screens the applicant's email and ID as a tool call, and attaches the result to the case.

Payment-ops automation

Agents screen counterparty wallets and websites before proposing or executing a payout.

Compliance copilots

Give an internal assistant the ability to answer "is this address sanctioned?" with a live check and the official source record.

Jurisdiction-aware agents

Agents call geolocate_ip and detect_vpn to factor location context into what they do next.

Scope and limitations

The MCP server is a transport for the same screening capabilities as the REST API — the same scope notes apply, plus some that are specific to agents:

Same data scope as the REST API

Tools check the US Treasury OFAC SDN list, curated country designations, the OpenSanctions PEP dataset (screen_pep), and IP signals. No adverse-media or blockchain-analytics tools.

Agents decide how to use results

A tool result is an input to the agent's reasoning. For critical paths, enforce deterministic checks in your application code as well — the REST API is the same data.

Review the trail

Every tool call is logged for your organization. Use that log to review what your agents actually screened.

Frequently asked questions

Does the MCP server use the same API credentials?

Yes. Authenticate with your existing CompliAPI API token as a bearer header. Tool calls are metered against your organization's quota exactly like REST requests.

Which tools does the server expose?

screen_crypto_address, screen_email, screen_website, screen_government_id, screen_country, screen_pep, detect_vpn and geolocate_ip — the full set of CompliAPI screening capabilities.

Which MCP clients are supported?

Any client that supports remote MCP servers over streamable HTTP. Add the server URL and an Authorization header to the client's MCP configuration.

Is the data different from the REST API?

No. The MCP tools call the same screening logic and the same OFAC SDN data, refreshed every 15 minutes, as the REST endpoints.

Should agents be the only place screening happens?

For critical flows, most teams also enforce checks deterministically in application code — the REST API exposes the same data with the same credentials, so the backstop is one request away.

Sanctions screening

OFAC API — screen against the official SDN list

Screen crypto wallets, emails, IDs and countries against the official US Treasury OFAC SDN list, refreshed every 15 minutes.

Sanctions screening

Wallet screening API for crypto compliance

Screen crypto wallets against OFAC and global sanctions lists in one GET request.

Use case

Compliance tools for AI agents

Give AI agents OFAC screening capabilities over MCP: seven sanctions-screening tools with the same API tokens, quotas and request logging as the REST API.

Technical documentation →From the blog: Role of AI in crypto-company compliance →

Give your agents OFAC screening tools

Get a free API key and connect the MCP server in minutes.