Decide on your screening approach: direct integration with the OFAC SDN list (downloaded as XML/CSV from the OFAC website) parsed and matched locally, or a commercial watchlist screening vendor (e.g., ComplyAdvantage, Dow Jones, LexisNexis WorldCompliance, or similar) that provides pre-normalized, multi-list coverage via API.
For vendor-based screening, submit the individual or entity name, date of birth, nationality, and any identifiers via the vendor's API at onboarding; parse the response for match results categorized by list (OFAC SDN, OFAC consolidated, PEP, adverse media, etc.).
Implement a fuzzy name-matching threshold; exact-match-only screening misses transliteration variants and is insufficient for compliance—consult your compliance officer on acceptable match score thresholds.
Route potential matches (above your threshold) to a compliance analyst for manual review; implement a clear match/false-positive/true-positive adjudication workflow and maintain records of decisions.
Set up ongoing monitoring (re-screening existing customers when watchlists update) in addition to point-in-time onboarding screening; most commercial vendors provide ongoing monitoring as a subscription feature.
Document your screening program, match criteria, and escalation procedures in your AML/BSA policy; regulators expect written procedures, not just technical controls.
Known gotchas
OFAC compliance is not optional for US persons and entities; failing to screen against the SDN list exposes your business to civil and criminal penalties regardless of whether a match is a true positive.
Name-only screening generates excessive false positives; adding date of birth and nationality dramatically improves precision—collect these fields during onboarding specifically to support screening.
Watchlist data changes frequently (new designations, updates, removals); if self-hosting the SDN list, automate daily refreshes; staleness of even a few days can create compliance gaps.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp