How the library works: one namespace, honest licensing, reproducible downloads, and a public update pipeline.
Every series has a stable id of the form source:series_key[:geography] — for example worldbank:NY.GDP.MKTP.CD:USA. The id is permanent, appears in every download, and resolves over the API.
Redistributed sources have licenses that permit re-hosting: their data is served from our store as citation-headed CSV, over the API, and in bundles. Metadata-only sources have licenses that forbid re-hosting: they are fully searchable and carry machine-readable metadata and pointers, but the data stays with the publisher — a direct request returns an honest HTTP 451 with the publisher's link. Nothing restricted is ever silently redistributed.
Bundles are snapshot-pinned: a bundle manifest records the snapshot date and the exact member series, so the same request reproduces the same data. Every CSV carries its license and producer-first citation in a comment header.
Sources are refreshed by a cadence-aware pipeline (daily, weekly, monthly, annual — matching each publisher's own schedule). Freshness is never fabricated: a series' date advances only when observations were actually fetched, and failures surface on the public status board rather than being hidden.
Series search is available in six languages (English, Arabic, Spanish, French, Russian, Chinese) using only the sources' official translations — titles are never machine-translated.
The free ElkassabgiData key works across the family — this library and HF Data Library (1-minute U.S. equity data). Get a key from the Download page.