Sanctions Check — OFAC SDN Screening, Pay-Per-Query

Pay-per-event — zero monthly fee, zero subscription. You pay for each query executed and each match found, nothing else. No seat licenses, no per-record data fees, no idle-month charges.

---

What it does

This actor searches the U.S. Treasury OFAC Specially Designated Nationals (SDN) list — the primary sanctions screening list used for U.S. compliance — by name, alias, address, or identifier. It has two modes:

• refresh — downloads the current OFAC SDN XML (~27 MB, ~19,000 entries) and builds a token-based inverted index in Apify's Key-Value Store: every name, alias, address city/country, and ID number is normalized and split into searchable word tokens, so partial-name queries resolve correctly against the full dataset instead of requiring an exact match.
• query — searches that index by name or identifier with substring/fuzzy scoring, and returns every matching SDN record: entity name, type (Individual/Entity/Vessel/Aircraft), sanctions programs, aliases, addresses, ID numbers, and remarks.

If you're looking for an OFAC SDN API without building your own XML parser and refresh pipeline, this gives you structured JSON matches for any name query, on demand or scheduled.

---

Use cases

• KYC / customer onboarding — screen a new customer, counterparty, or beneficial owner's name against the SDN list before account opening.
• Supplier / vendor due diligence — check suppliers, contractors, or M&A targets for sanctions exposure before signing.
• AI agent compliance tool — give an LLM agent a callable sanctions-screening function via the Apify MCP server (see below).
• Journalist / researcher due diligence — quick name checks against public sanctions data.
• LatAm due-diligence profiles — pair with cuit-lookup (AFIP/CUIT identity data) and licitaciones-feed (public tender awards) to build a combined counterparty profile.

---

Input

{ "mode": "refresh" }

{ "mode": "query", "query": "MADURO", "searchType": "all", "maxResults": 10, "minScore": 0.5 }

Field	Type	Default	Description
mode	refresh | query	query	refresh downloads and rebuilds the index (run this at least once before querying). query searches the existing index.
query	string	—	Name or identifier to search (case-insensitive). Required in query mode.
searchType	name | all	all	name = match entity name only. all = also match addresses, IDs, and aliases.
maxResults	integer	20	Maximum matches to return (1–100).
minScore	number	0.5	Fuzzy match threshold: 1.0 = exact, 0.7 = substring, 0.5 = partial word.

Workflow: run once with { "mode": "refresh" } (~30–45s, downloads + indexes the full list), then run { "mode": "query", "query": "..." } for each name you want to screen. Schedule refresh daily (or at whatever cadence matches your compliance policy) to keep the index current — see Freshness below.

---

Output

Each match is one dataset row. Example — querying "MADURO" against a freshly refreshed index:

{
  "query": "MADURO",
  "name": "MADURO MOROS, Nicolas",
  "sdnType": "Individual",
  "programs": ["VENEZUELA", "IRAN-CON-ARMS-EO"],
  "remarks": "President of Venezuela.",
  "addresses": [
    { "city": "Caracas", "country": "Venezuela", "address": "" }
  ],
  "ids": [],
  "aliases": [
    { "name": "MADURO MOROS, Nicolas Maduro", "type": "aka" }
  ],
  "source": "OFAC-SDN",
  "sourceUrl": "https://sanctionslistservice.ofac.treas.gov/api/publicationpreview/exports/sdn.xml?uid=22790",
  "lastUpdated": "2026-07-24T00:00:00.000Z",
  "searchedAt": "2026-07-24T13:05:00.000Z"
}

Field	Type	Notes
query	string	The search term you submitted
name	string	Matched entity's display name (lastName, firstName for individuals)
sdnType	string	Individual, Entity, Vessel, or Aircraft
programs	string[]	Sanctions program codes (e.g. VENEZUELA, SDGT, KINGPIN)
remarks	string|null	OFAC's free-text remarks field, when present
addresses	array	City/country/address strings from the record
ids	array	Passport / national ID / other identifiers, with type and issuing country
aliases	array	Known AKA names, with alias type
source	string	Always "OFAC-SDN" in v1
sourceUrl	string	Direct link back to the source dataset for this record
lastUpdated	string	Timestamp of the OFAC file this index was built from
searchedAt	string	Timestamp this query ran

A query that finds zero matches still runs and charges the sanction-screened event (a clean "not found" is a valid compliance result, not an error) — it just pushes no rows and charges no sanction-match events.

---

Pricing (Pay-Per-Event)

Event	What triggers it	Price
sanction-screened	Per query execution (mode: "query"), even if zero matches	$0.003
sanction-match	Per match found and returned	$0.01
Actor start	One-time per run. Apify charges this event once per GB of run memory (minimum 1); this actor's default memory is 2048 MB (2 GB), so a standard run charges 2 actor-start events at $0.0005 each	$0.001 total/run

Each query is one search term per run (the input takes a single query string, not a batch array) — screening multiple names means multiple runs, each with its own actor-start + sanction-screened charge.

Real-world cost examples:

Scenario	Runs	Matches	Est. cost
Clean name, no hits (typical KYC check)	1	0	~$0.004
Common surname, several hits	1	4	~$0.044
Bulk screen, 50 names, ~10% hit rate	50	5	~$0.25

Measured platform cost: ~$0.0004–0.0005 per query in Apify compute-unit terms (validated build 0.1.4, platform runs ZmZvQSUWXes0L1dH8 and 4dMt9h3CY1CXRZZxn) — at $0.003/query that's roughly 83–87% gross margin before the platform's own PPE cut. refresh runs (index rebuild) are not separately billed beyond the actor-start charge; they cost the developer $0.03/run in compute ($1/month at daily cadence) and are intended to run on a schedule, not per-customer-query.

---

Use with AI agents

This actor is available as a tool for AI assistants via the Apify MCP server. Agents can call it to screen a name against the OFAC SDN list directly — no Apify console needed, just a single function call.

MCP setup

{
  "mcpServers": {
    "apify": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.apify.com?tools=mfapitools/sanctions-check",
        "--header",
        "Authorization: Bearer <YOUR_API_TOKEN>"
      ]
    }
  }
}

Example agent prompts

• "Check if 'Nicolas Maduro' appears on the OFAC SDN sanctions list"
• "Screen this list of 20 supplier names against U.S. sanctions before we onboard them"
• "Is this company or any of its known aliases sanctioned?"

The actor returns structured JSON with the same fields shown above, so an agent can reason over match strength (programs, aliases, remarks) without parsing raw XML.

---

Honest limitations

• NOT a substitute for professional compliance software. This is a name-search tool over a public dataset. It does not perform risk scoring, PEP screening, adverse media checks, or transactional/payment screening.
• OFAC SDN only in v1. UN, EU, and UK sanctions lists are not covered. If your compliance policy requires multi-list screening, this covers only the U.S. SDN list.
• Query candidate retrieval requires a whole matching word (token), not an arbitrary substring. A query for "MADURO" finds records because MADURO is indexed as a complete token. A query for "MADUR" (a sub-word fragment that is never itself a complete word in the index) will not retrieve any candidates — there's no entry for the fragment "MADUR", only for the full word "MADURO". Once a candidate record IS retrieved via a full-token hit, the existing substring/fuzzy scoring still evaluates the WHOLE query string against that record's fields for ranking — so "MADURO MOR" still scores well against "MADURO MOROS, Nicolas" (retrieved via the "MADURO" token, then scored as a partial-word match). Very short typos inside a single word (e.g. "MADURRO") are not corrected — there is no edit-distance/phonetic matching.
• Multi-word queries use UNION, not intersection, of per-word candidates — chosen deliberately to favor recall for a screening tool (a record shouldn't be silently dropped just because it only matched one of several query words). This means a two-word query where only one word is real can still surface a match; the score/ranking (not the candidate set) is what tells you how strong the match is.
• Freshness depends on how often you run refresh. OFAC updates the SDN list as needed (sometimes daily). This actor does not auto-refresh — schedule refresh runs at a cadence that matches your compliance requirements, and treat lastUpdated on each result as the authoritative "as of" date.
• ~19K entries — a small, focused dataset. The value here is search/index convenience and structured output over the SDN list, not data volume.
• No fuzzy phonetic matching (e.g. Soundex). Name variant spelling that shares no common whole-word token with the indexed name will not be found.

---

Compliance

• Public-domain government data. The OFAC SDN list is published by the U.S. Department of the Treasury and is explicitly public domain — no license, login, or API key is required to access it.
• Source: https://sanctionslistservice.ofac.treas.gov/api/publicationpreview/exports/sdn.xml, the same file OFAC's own compliance tools and every commercial screening vendor ultimately index.
• Update cadence: OFAC republishes the SDN list as needed (frequently, sometimes daily, whenever designations change). This actor reflects whatever version was present the last time refresh was run — it does not poll OFAC continuously. Run refresh on a schedule matching your risk tolerance.
• No sensitive personal data beyond the source record. Output fields (name, aliases, addresses, ID numbers, remarks) mirror exactly what OFAC publishes in the SDN file — nothing is enriched, inferred, or sourced elsewhere.

---

Changelog

See ./CHANGELOG.md.