[claude-tmgbedu]

Why

#307, #309 and #310 were merged into their stacked parent branches (task/cache-memo and task/cache-lock), so Cache.lock, Cache.tags and the Redis lock never reached feature/caches. Only #306 (Cache.memo) did.

What

• Merges origin/task/cache-lock, which holds Cache.lock(): atomic locks for array, file and database stores #307, Cache.tags(): namespaced tagged cache with tagged flush #309 and Cache.lock(): Redis lock #310, plus the task/cache-tags, task/cache-redis-lock and task/cache-memo heads into feature/caches. All merges were clean, and no code was changed.
• Folds the cache ADRs into one: docs/adr/029-cache-module.md now covers the store module, memo, locks and tags. It removes 030-cache-memo.md, 031-cache-locks.md and 034-cache-tags.md and the stale "Out of scope: memo, tags, locks" line, and leaves a single 029 row in the index.

Content verification (against this branch)

Branch	Non-merge commits missing	Notes
task/cache-config, -core, -remember-decorator, -provider-facade	0	Only the stacked-PR merge commits are not ancestors
task/cache-file-store, -database-store, -redis	0	Ancestors
task/cache-memo, -lock, -tags, -redis-lock	0	Ancestors after this merge

I also checked each PR's added lines (#305, #306, #309 and #310) against this tree: every line is present. The misses in #296–#303 and #307 are all lines that later commits deliberately rewrote, such as the unittest test rewrite, the DB encode() helper, the Redis config and ADR text updates.

Validation

• Full suite: 3287 passed, 10 skipped.
• Cache: 241 passed.
• ruff and basedpyright clean.

Supersedes #311.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!