Introducing the LEI Search MCP Server: Plug GLEIF Plus Risk Signals Into Any AI Agent

Why a Second MCP Endpoint

GLEIF's official AI Connect servers (covered in our earlier post) expose the raw Global LEI Index. The LEI Search MCP server sits one layer higher: it composes GLEIF data with the consolidated sanctions index, the deterministic AMLA risk rule set, and the Beneficial Ownership Data Standard (BODS) bundle exporter we ship on every entity page. The result is a single tool surface that an agent can use to answer compliance and due diligence questions end to end.

Endpoint

• URL: https://leisearch.com/mcp
• Transport: Streamable HTTP (MCP spec)
• Auth: public for read tools, Pro API key required for get_evidence_bundle
• Service descriptor: GET https://leisearch.com/mcp/info

The Five Tools

1. lookup_lei

Returns the full GLEIF record for a 20-character LEI: legal name, addresses, jurisdiction, status, next renewal date, managing LOU, BIC links, and canonical URLs on both leisearch.com and search.gleif.org.

2. search_entity

Free-text search across GLEIF legal names with an optional ISO 3166-1 alpha-2 country filter. Returns up to ten matches with LEI codes ready for downstream calls.

3. get_risk_signals

Computes our deterministic, AMLA-aligned risk chips on demand. Each result includes the rule id, severity (danger / warn / info), a short methodology, the AMLA reference, and the supporting evidence JSON. Covers sanctions exposure (direct and via parent chain), trust or arrangement legal forms, multi-layer ownership, jurisdiction risk, lapsed or pending registrations, and LOU fragmentation.

4. get_relationships

Returns GLEIF Level 2 corporate relationships: direct parent LEI, ultimate parent LEI, and up to 50 direct children with legal names and countries. Pair it with lookup_lei to walk an ownership tree.

5. get_evidence_bundle (Pro)

Returns a Beneficial Ownership Data Standard (BODS) v0.4 statement set for the entity and its immediate parent / child chain. Suitable for ingestion into ownership graph tools or for attaching to compliance files. Requires a Pro plan API key passed as the apiKey argument.

Example Conversation

1. User asks: "Is BNP Paribas exposed to any sanctions risk?"
2. Agent calls search_entity with name: "BNP Paribas", picks the parent LEI.
3. Agent calls get_risk_signals with that LEI, receives the chip list with evidence references.
4. Agent replies with a plain-English summary citing the matching rule ids and pointing at the entity page on leisearch.com for the full audit trail.

How It Is Built

The endpoint runs on a Lovable Cloud Edge Function using mcp-lite with the Streamable HTTP transport, wrapped in a Hono router. lookup_lei, search_entity, and get_relationships call the public GLEIF API directly. get_risk_signals and get_evidence_bundle proxy our existing risk-signals and bods-bundle edge functions so logic stays in one place. Tool calls are streamed back over MCP without any client-side glue.

Data Sources and License

All upstream data is public: GLEIF (CC0-1.0), OpenSanctions consolidated lists (CC BY-NC 4.0 for the underlying datasets, used for risk signals only, not redistributed), and BODS v0.4 statement schema from Open Ownership. Pro API keys are managed under the standard LEI Search subscription terms.

What's Next

We plan to add screen_sanctions (single-entity fuzzy screening), get_renewal_status (compliance tracker integration), and a get_cross_references tool that surfaces Wikidata, OpenAleph, and EU TED hits in one call. If you have a specific workflow you want exposed as an MCP tool, the contact form on the site is the fastest path.