docs(agents): record 0.22.0 release, #892/#894 landings, 2026-08-28 tracker items

[pmrv]

Scheduled AGENTS.md audit. Changes since the 2026-08-27 pass:

• Release-state claims updated: PR feat(caches)!: seal call identity before the function body runs #793 (two-phase save) and PR feat(storage): atomic renames instead of file locks in pickle-family backends #843 (lock-free pickle-family backends) shipped in 0.22.0 on 2026-08-28 (release PR chore(main): release 0.22.0 #837); the three "unreleased as of 2026-08-27" qualifiers are gone. Noted that fix(query) fix(query): raise instead of silently picking an arbitrary latest()/oldest() #894 is an ancestor of the 0.22.0 tag (verified via git tag --contains c82c691) even though the generated changelog entry omits it.
• PR fix(query): raise instead of silently picking an arbitrary latest()/oldest() #894 (latest()/oldest() raise ValueError when no matching call carries Runtime metadata, IndexError on empty), PR test: cover Cache.gc()'s mid-sweep races; collapse the 21x21 digest product #892 (gc mid-sweep guards + 441-case digest product collapse, suite 1738 → 1299), and docs-audit PRs docs(usage): fix inaccurate/misleading claims in tldr, helpers, query #873/docs(installation,parallel_execution): fix quoting bug, trim duplication, close gaps #874 moved from the in-flight section to decisions landed; the in-flight section is now a one-liner pointing at the five PRs still open (fix(state): serialise BoundWrapper.func by value when not importable by reference #887, fix(benchmarks): don't measure hit-phase timing against evicted keys #804, feat(storage): cache filesystem paths by content #797, fix(storage): cache open HDF5 read handles for multi-bag bag files #786, remove isolate from fleche decorator #523), each covered under its theme.
• New tracker items recorded: the 2026-08-28 refactor cohort Refactor: tests/strategies.py's dataclasses() strategy instantiates via cls(*fields) — unpacks dict keys, not values #899–Refactor: Runtime._keys unconditionally advertises cputime/systime even though pre/post omit them when resource is unavailable #902, bug put() dies on file-lock Timeout while get() recovers; content-addressed dedup concentrates writers onto one lock #903 (put() filelock.Timeout fatality on ≤0.21.2, dedup-concentrated lock contention; resolved for pickle-family by 0.22.0's atomic rename, bagofholding multi-bag still exposed per bagofholding_file prefix-bag writes: filelock.Timeout storms at the 1.0 s default under same-bag concurrency (measured; not a deadlock) #893), and the wrap_executor: cache-hit Future is always stdlib Future, even for executors that return a different Future subtype #895 resolution (documented limitation per the 2026-08-28 maintainer comment). Refactor: pypi-publish.yml and release-please.yml duplicate the whole build+publish sequence — and can both fire on the same release #900 cross-referenced at the pypi-publish.yml line in the workflows list.
• The bagofholding_file prefix-bag writes: filelock.Timeout storms at the 1.0 s default under same-bag concurrency (measured; not a deadlock) #893 bullet's lock_timeout mitigation sentence rephrased to the current state: the FutureWarning drop covers only pickle-family configs, a bagofholding values.lock_timeout still works on 0.22.0 (checked against config.py:466).
• USAGE.md: query section states the latest()/oldest() ValueError/IndexError contract (verified against query.py:149-188).

landau's AGENTS.md was audited in the same pass and left unchanged — no code or tracker movement since its 2026-08-24 update (only docs commits on main; issues #423–#425 and PRs #422/#414/#395/#394/#391 unchanged).

🤖 Generated with Claude Code

https://claude.ai/code/session_014myhh6xKWC2VhQeB92mGSn

---

Generated by Claude Code