canonical-catalog-expansion
Grow the canonical data catalog without degrading schema quality.
Procedure
- Read
docs/canonical-data-catalog.md, the catalog ADR,docs/governance/rfc-process.md, data-storage boundary, and naming guide. - Use
claw collections list,claw collections <collection> schema, andclaw db <collection> list|querywhen available. - Decide whether the entity is built-in canonical or belongs in a custom database.
- For promotion to canonical type or canonical user-profile attribute, require the RFC process record, public review link, and maintainer sign-off before treating the shape as accepted canon.
- Add purpose, evidence tags, sparse optional fields, semantic relation fields, aliases, and migration/debt notes.
- Keep user-facing structured records in the main database unless a sidecar reason is technical and explicit.
- Update docs, tests, generated catalog coverage, and CLI/schema discovery together.
Constraints
- A business category alone is not a reason for a separate database.
- Separate sidecars for volume, churn, blobs, sync complexity, logs, caches, encrypted vaults, or reconstructable indexes.
- Do not let Clawix define a parallel canonical schema.