drugref.org is a public-good drug-information service: a registry of active drug moieties, each with an immortal identity and transparent, append-only links to the world's open drug vocabularies. Any EHR, pharmacy system, prescribing tool or research pipeline can consume it on equal footing — free software, free data, reproducible from source.
Reliable drug reference data — what a substance is, how it is classified, what it interacts with — sits behind proprietary licences in most of the world. That locks safe prescribing support to whoever can pay, and locks health-IT projects into vendors. drugref.org takes the opposite stance: the identity and knowledge layer of medicines is a commons, built entirely from public-domain and openly licensed sources, and published under a licence that keeps it open for everyone, forever.
This is a resuscitation of the original drugref.org project, rebuilt from the ground up as drugref v2 with modern identifiers, reproducible ingestion, and a design hardened for decades of use.
Every active drug moiety gets a stable UUID that never changes and is never reused — minted deterministically, so independent installations agree on identity without coordination.
Append-only, auditable links from each moiety to UNII, InChIKey, CAS, ChEBI, RxNorm and more. Corrections supersede — they never silently overwrite history.
Source-neutral drug class registry populated from MED-RT (mechanism of action, physiologic effect) and MeSH pharmacologic actions.
Class-level contraindication data from MED-RT, expanded to candidate drug–drug pairs — the advisory foundation for interaction checking, clearly tiered by evidence.
An authenticated desktop workflow keeps research notes, immutable clinical revisions and detached Ed25519 signatures distinct. Private signing keys stay on the reviewer's device, and corrections preserve history.
Missing classifications, unresolved source objects and clinical questions remain queryable worklists instead of disappearing as silent absence. Coverage improves in public, one reviewed gap at a time.
Two tiers are planned: the global tier (jurisdiction-independent substance identity, chemistry, classes, interactions — being built first, in this repository) and a later local tier for country-specific packaging, product and subsidy data.
Every fact in drugref traces to a public-domain or openly licensed upstream release, and the entire database can be rebuilt from those sources by anyone:
Licensing is a hard rule, not an afterthought: only AGPL-compatible sources are ever bundled. Licence-encumbered national sources can attach to a local installation as separate plug-ins, but never contaminate the commons.
drugref.org is stewarded alongside the Cairn EHR project, but it is not a Cairn component. Cairn consumes drugref through the same public API offered to everyone — the steward has no privileged access. drugref data is advisory reference information: it informs clinical software, but no system is ever required to route its records through it.
The project is developed in the open at github.com/cairn-ehr/drugref — code, public documentation, design specifications, roadmap and issues. Contributions are welcome: source-licence review, ingestion of additional open vocabularies, clinical review, API design, and national local-tier pilots are all on the horizon. Start with the reviewer manual or developer guide.
All code is licensed AGPL-3.0; bundled reference data retains its upstream open licences (see the NOTICE file).