For developers

Sanctions Screening API and MCP Server for AI Assistants

Our REST API and MCP server screen names against 10 official sanctions lists. A free key with 100 credits comes by email from the form on this page.

Get a free API key MCP setup page

100 free credits
No subscription
OpenAPI and MCP server

Connect in 3 steps

Get your API key

Type your email address in the form below and the key arrives by email.

Screen names with the REST API

Send the names in one POST request with the key in the Authorization header.

curl -X POST https://sanctionlistcheck.com/v1/screen \
  -H "Authorization: Bearer $SLC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"names":[{"name":"Acme Trading"}]}'

Connect your AI assistant

Add the MCP server to Claude Code with one command; the setup page shows the other clients.

claude mcp add --transport http sanctionlistcheck https://sanctionlistcheck.com/mcp \
  --header "Authorization: Bearer $SLC_API_KEY"

Get your key

Type an email address and the key is sent to it, never shown on the page. A new key has 100 free credits and one credit screens one name.

One API key per email address, and at most 3 new keys from one network in 24 hours.

The email address is used only to send the key and account notices about the key. Privacy Policy

A script or a terminal can ask for a key with this command; the key still arrives by email.

curl -X POST https://sanctionlistcheck.com/v1/keys -H "Content-Type: application/json" -d '{"email":"you@example.com"}'

REST API

One POST request screens 1 to 100 names and answers JSON. Each screening creates a record with a record id, the snapshot version, the data-as-of date of each list and the engine version. The same key can fetch, list and delete its records; records are kept 90 days by default and each key can set its own retention from 1 to 365 days.

MCP server

AI assistants such as Claude Code, Claude Desktop and Cursor connect to the MCP server and call its tools to screen names, fetch a record, list the source lists and check the balance. It uses the same key and credits as the REST API.

The MCP setup page

Where data is processed

Stored data has the Cloudflare location hint EEUR, a placement preference, not a legal jurisdiction. Every request passes through Cloudflare's global edge network, which can be outside the EU. Name matching normally runs on a matching server in Helsinki, Finland, hosted by Hetzner; if that server does not answer, returns an error, or we switch it off, matching runs in Cloudflare's network, which can be outside the EU. Customer data is never used to train AI models.

What one screening returns

POST /v1/screen answers JSON with these fields.

Field	Meaning
`record_id`	The id of the stored record of this screening.
`record_url`	The URL of that record. Fetch it with the same key; its results are kept until payload_expires_at.
`screened_at`	When the screening ran, in UTC.
`snapshot_version`	The version of the list snapshot that was used.
`engine_version`	The version of the matching engine.
`sources`	Each source list with its name and the date of its list data (data_as_of).
`warnings`	Notes about this screening, for example when the list dates are not available (data_as_of is then null).
`request_hash`	A hash of the request. Without an Idempotency-Key, the same request is not charged twice if you repeat it within 1 hour. With the same Idempotency-Key, it is not charged twice within 24 hours.
`threshold`	The lowest score that is returned as a possible match.
`results`	One item per input name: the input and its hits. Each hit has the list, the matched name, a score, a band (high_confidence or possible_match) and the list entry data.
`credits_charged`	The credits this call used.
`credits_remaining`	The credits left on the key.
`payload_expires_at`	When the stored results of this record are deleted.
`payload_status`	available when the results are in the answer; expired or missing when results is null; none when the record is pending or failed, so there are no results (GET /v1/records).
`disclaimer`	The terms and limits of the results: provided as-is, your own decisions, liability capped at fees paid, not legal advice, and a note on names written in other scripts.

Credit packs

More credits come as one-time packs with no subscription, credits do not expire while the key is active. The checkout link for your key comes from the packs endpoint or from the API answer when the credits run out.

Pack	Credits	Price
Free key	100	Free
Starter	500	€15
Growth	2,500	€60
Scale	10,000	€200

Credit packs are charged in EUR.

curl https://sanctionlistcheck.com/v1/packs -H "Authorization: Bearer $SLC_API_KEY"

Reference links

Questions and answers

What happens when the free credits run out?

When your 100 free credits run out, you can buy one-time credit packs. The API answer will give you a buy link for more credits.

Which AI assistants can connect?

AI assistants such as Claude Code, Claude Desktop and Cursor can connect to the MCP server. The MCP setup page shows the setup for each client.

What does a hit mean?

A hit is a candidate for human review, not a finding. No hits does not mean a name is cleared. The results are a screening aid, not a clearance.

Is customer data used to train AI models?

Customer data is never used to train AI models.

The results are a screening aid, not a clearance and not a compliance verdict. A hit is a candidate for human review, no hits does not mean a name is cleared, and the customer stays responsible for their own compliance decisions.

More Guides