FerroPIX
The patient index FerroEHR asks for the pseudonym an
EHR_STATUS subject carries, so identifying data
stays outside the clinical record.
openEHR separates clinical knowledge from software: applications store and query structured health records through a vendor-neutral REST API and the Archetype Query Language. FerroEHR implements that standard natively in Rust: headless, API-first, one self-contained binary on PostgreSQL 18.
A server image, a viewer image and a Helm chart on GHCR, built at SLSA Build Level 3, with no JVM and no language runtime, and the openEHR model crates on crates.io.
Generated from the official machine-readable models, so a specification update is a regeneration, and every service is reached over the openEHR REST API.
EHR, EHR_STATUS, COMPOSITION, DIRECTORY, CONTRIBUTION, query, template and admin resources, in canonical JSON and XML on the wire.
AQL 1.1
Typed path analysis over a specification-generated Reference
Model, compiled to SQL, including ALL_VERSIONS and
stored queries.
Contribution-atomic commits, indelible version history with database-enforced non-overlap, logical delete, attestations and per-version digital signatures.
OPT 1.4Template ingestion, WebTemplate generation and deep composition validation against archetype constraints and terminology bindings.
OIDC, SMART and IHE ATNAOIDC authentication, RBAC and ABAC authorization, SMART App Launch and an IHE ATNA audit trail. One instance per organisation.
Beyond the coreA change-event feed, EHR Extract messaging, demographics, a FHIR terminology server and S3 multimedia storage connect openEHR to the systems around it.
Veredictum, a separate openEHR conformance instrument, runs its whole catalogue against a live FerroEHR on every push to main. One run against the released specification is committed as the record, and every number below is read from it.
The committed record: FerroEHR 5.6.0-rc10, veredictum 0.4.1, released profile, 2026-11-17.
ferroehr-viewer is a web UI for any ITS-REST 1.1.0
CDR. It goes through the public API and never the database, so
what you see is exactly what the API serves: queries, templates,
EHRs, parties, terminology and the audit log.
The server is one self-contained binary on a minimal distroless image, beside PostgreSQL 18. The viewer is its own image.
docker-compose.yml ships with every release:
docker compose up starts the server and PostgreSQL,
and the viewer profile adds the viewer.
ghcr.io/ferrohealth/ferroehr and its PostgreSQL
image, for amd64 and arm64, each with SLSA provenance and an SBOM
you verify before you run it.
oci://ghcr.io/ferrohealth/charts/ferroehr runs the
server on Kubernetes, with the cluster hardening guide beside it.
The FerroHEALTH products reach each other only over published protocols. FerroEHR is where the openEHR record lives.
The patient index FerroEHR asks for the pseudonym an
EHR_STATUS subject carries, so identifying data
stays outside the clinical record.
Coded text is validated on a FHIR terminology server such as FerroTERM, and FHIR mapping and broker delivery live in FerroBRIDGE, a separate server.
The federation gateway sends each member CDR ordinary AQL scoped to its own EHR ids, and FerroEHR answers as a federation node.
Why FerroEHR exists, the openEHR concepts it implements, its measured conformance and performance, and the licence.
The REST resources, AQL, templates and validation, and the services beyond the core.
Installation and configuration, cluster hardening, security and audit, the viewer, and day-to-day operations.
The tracker is the worklist, every check is a committed script you can run, and a change cites the specification section it implements.