@absolutejs/dataset-gleif

A @absolutejs/discover DatasetSource over GLEIF — the Global Legal Entity Identifier registry, published as open CC0 data.

import { gleifSource } from "@absolutejs/dataset-gleif";
import { discoverContacts } from "@absolutejs/discover";

const gleif = gleifSource();

await gleif.findCompany({ name: "Stripe, Inc." });
// → { name: "STRIPE, INC.", country: "US", registryId: "<LEI>", source: "gleif" }

// Or hand it to discover as a seed source consulted before LLM/web:
await discoverContacts({ company: "Acme" }, { sources: [gleif], extract });

What it does — and doesn't

GLEIF is legal entities, not people, so this adapter implements findCompany only. It canonicalizes a name to its official legal entity and returns the LEI (registryId) — the key to follow up for parent/subsidiary relationships in the GLEIF graph — plus the registered country.

Honest limits: GLEIF carries no domains, so matching is by name and is inherently ambiguous — a fulltext "Stripe" can surface an unrelated "Stripe B.V.". Treat a hit as "a legal entity by this name", not a certainty. It also has no industry, headcount, or contacts — pair it with @absolutejs/discover (people) and @absolutejs/enrich (emails). Defaults to ACTIVE entities only.

Offline snapshot — no rate limits, no network

For high-volume or offline use, build a local SQLite index from GLEIF's full Golden Copy dump (~2.7M entities) and point the adapter at it:

1. Download the Golden Copy (LEI2, CSV or .zip) from https://goldencopy.gleif.org (it's CC0). Their bulk host WAF-blocks many data-center IPs, so fetch it from a browser / allowed network — this tool does not download it for you.
2. Build the snapshot:

   bunx gleif-snapshot golden-copy.zip gleif-snapshot.sqlite
   # or: bun run node_modules/@absolutejs/dataset-gleif/dist/snapshot.js <file> [out]

3. Use it:

   gleifSource({ snapshotPath: "gleif-snapshot.sqlite" });

   findCompany resolves from the snapshot — instant, offline, zero rate limits — falling back to the live API only on a miss. Re-run the generator to refresh (GLEIF republishes 3×/day). Snapshot reads use bun:sqlite (loaded lazily; live-only use needs nothing).

Apache-2.0. Pure importer of public CC0 data — the self-collected, self-healing dataset is deliberately not part of any adapter.