mcp-sanctions - Watchlist Screening for KYC and AML¶
Stdio MCP server (PyPI) - screens persons and organizations against public watchlists (Rosfinmonitoring, OFAC SDN, EU consolidated, UK OFSI, UN Security Council) with neutral match output: match level, list version and check date in every answer, plus a disclaimer. Manual verification remains required.
Spec Block¶
| Field | Value |
|---|---|
| Server name | atomno-mcp-sanctions |
| Repo | github.com/atomno-mcp/mcp-sanctions |
| PyPI | atomno-mcp-sanctions v0.1.1 |
| Transport | stdio (pipx or uvx), hosted backend for data |
| Backend | https://api.atomno-mcp.ru/sanctions |
| Auth | Pro key required, sent as X-API-Key (env MCP_SANCTIONS_API_KEY) |
| License | MIT |
| Stars | 0 (new repo) |
Why This Matters for Operators¶
Screening a counterparty against five watchlists in one call, with an explicit neutral match level instead of a binary verdict, is exactly the output a compliance workflow needs: the tool reports the match against a dated list version and leaves the decision to a human. It is built to be an input to KYC/AML processes, not a replacement for them.
Tools & Capabilities (4 tools)¶
| Tool | Description |
|---|---|
| screen_person | Screen a natural person by full name, optionally with date of birth |
| screen_entity | Screen an organization by name, INN or OGRN |
| screen_inn | Quick lookup by Russian tax identification number |
| list_watchlists | Metadata on supported lists and their update dates |
Installation¶
pipx install atomno-mcp-sanctions
# or: uvx atomno-mcp-sanctions
Cursor or Claude Desktop (mcp.json):
{
"mcpServers": {
"sanctions": {
"command": "uvx",
"args": ["atomno-mcp-sanctions"],
"env": { "MCP_SANCTIONS_API_KEY": "<your-pro-key>" }
}
}
}
Configuration¶
- MCP_SANCTIONS_API_KEY - Pro key from atomno-mcp.ru/pricing, sent as the X-API-Key header (required)
- MCP_SANCTIONS_API_BASE - hosted backend URL (default https://api.atomno-mcp.ru/sanctions)
- MCP_SANCTIONS_TIMEOUT - HTTP timeout in seconds (default 30)
Business Relevance¶
Useful for: compliance officers screening counterparties across five jurisdictions at once, KYC onboarding checks with auditable list versions and dates, and due-diligence workflows where the answer must state its source rather than assert a verdict.
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.
Limitations¶
The Central Bank of Russia denylist is not supported (not public). The tool explicitly does not make judgements and is not a basis for decisions on its own; every answer carries a disclaimer that data may age. Pro key required for the hosted data backend.