About the registry

The Assessment Metadata Registry is a general, version-controlled catalog of state-assessment and accountability-system metadata. Its purpose is to make the facts that govern how assessment results are interpreted — scales, achievement levels, cutscores, vertical-scaling flags, comparability caveats, and accountability targets — auditable, open, and reproducible.

How it is built

The registry follows a strict three-tier model:

Tier What Canonical?
A — Authored JSON sidecars under metadata/<jurisdiction>/<system>/*.json Yes — the source of truth
B — Derived index, per-record bundles, changelog, SQLite, SHA-stamped manifest No — regenerated by amrr::build_registry()
C — Consumed the amrr R package, this catalog, static bundles No

This site is Tier C presentation: every page you see is rendered from the derived build/** JSON, which is itself derived from the authored sidecars. Nothing here is authored, and nothing here can be edited — it is a read-only lens on the data.

Reproducibility

Metadata is versioned by year (the sidecar) and by Git commit (the content). Each derived bundle carries a _registry stamp — schema_version, git_sha, dirty, built_at — so any past state can be reconstructed by checking out the recorded commit SHA. The catalog you are viewing was built from the commit shown in the footer.

Fetching the data

The machine-readable JSON is published alongside this site and is the canonical fetch target for programmatic consumers:

  • index.json — one row per jurisdiction × system × year × content area
  • changelog.json — year-over-year changes
  • manifest.json — every artifact with a sha256
  • dist/<JUR>.json and dist/<JUR>/<system>.json — full record bundles

The amrr R package consumes these, pinning the registry commit SHA it resolved against.