staleness-check

Check what governance artifacts went stale based on what was edited this session.

This is the TEMPORAL governance check — it answers: "What docs did my changes just make stale?" Run it before declaring a task complete.

Steps

1. Read .claude/memory/file_edit_log.yaml to find files edited today.

2. For each edited file, check ripple effects against the project's governance layer:

   If a file in lib/database/, models/, tables/, schema/, or migrations/ was edited (database/schema-touching code):

   • Is the schema doc current? (knowledge/schema.{yaml,md} or whatever your project uses)
   • Was a new column added but FK targets/indexes/constraints not documented?
   • If the table is claimed by any feature_flows/*.yaml, is that flow's participating_files and invariants still accurate?

   If a service/handler/api/endpoint file was edited:

   • If it's a critical-system service (auth, encryption, sync, payments, broadcast, or any flow tagged critical in your project's feature flow inventory), check the matching feature_flows/{system}_flow.yaml "last_updated" date — is it today? Are participating_files and declared_dependencies current?

   If a new screen / view / route was added:

   • Is it claimed by the relevant feature flow's participating_files?
   • If the screen lives in a critical user journey, is there a flow doc for it at all?

   If any feature flow was edited:

   • Is the "last_updated" date today?
   • Do declared_dependencies between flows still resolve to existing flow files?
   • If you removed a participating_files entry, did you confirm the file is genuinely no longer part of the flow (vs. just renamed)?

   If a Postgres / Supabase function / Edge Function / API route source was edited:

   • Is the function listed in knowledge/tech_stack.yaml (or your equivalent) under the relevant section?
   • Is the relevant feature flow's participating_files updated?

3. Check feature flow freshness overall — flag any flow whose last_updated is >7 days ago AND whose participating_files were touched this session.

4. If your project has an automated schema-drift detector (e.g., scripts/check_schema_drift.py), check its last run from scripts/RUN_LOG.yaml. If >7 days old AND any schema-touching file was edited this session, recommend running /check-drift (or equivalent).

5. Report:

   • Current — files whose docs/flows are accurate
   • Stale — docs/flows that should be updated as a consequence of the changes
   • Missing — code paths edited that have no corresponding documentation