Repository navigation
release: 4.23.0 - #609
release: 4.23.0#609
Conversation
|
ZETETIC-REVIEW: REQUEST_CHANGES Stakes: High (release surface, public manifests, marketplace delivery). Reviewed at 1bc073c in a detached worktree at Blocking: the zetetic-team-subagents pin advertises 2.42.0 while fetching the 2.41.0 tree
The entry as it stands on the branch: "source": {
"source": "github",
"repo": "cdeust/zetetic-team-subagents",
"ref": "v2.41.0",
"sha": "77b4f78503e6842928039b75440e20ea670584c8"
},
"version": "2.42.0",The sha is the 2.41.0 commit, not the 2.42.0 one: Every install resolving this entry gets The established convention moves all three fields in one commit. The last delivery, 8d71210 Required change, either one:
Why no job caught it: Everything else measured, and it holdsShape against the previous cut. Gates:
The NOTICE is the registry lagging the repo, resolved by the tag push. The failure is the self-pin naming a version whose tag does not exist yet, which is the correct state of a release branch and is exactly how #570 behaved: its uv.lock. Only the project's own entry moved, CHANGELOG. The diff is two added lines placing Release notes. The premise in the review brief does not match the workflow: README. CI. Unblocking this is two field edits in one file: |
|
ZETETIC-REVIEW: APPROVE This supersedes the round-1 comment at 1bc073c (REQUEST_CHANGES, #609 (comment)). Round 1's single blocking finding was that Diff since 1bc073c: Only those two fields changed, in one file. Nothing else moved. The sha matches what the v2.42.0 tag resolves to: The tag ref returns an annotated tag object; dereferencing that object gives the commit sha the PR pins, and it matches. The release exists. Pin checker, run at 1a8adf6: This is exactly the one allowed failure round 1 recorded, plus the same registry NOTICE, and the zetetic-team-subagents pin no longer appears in the output at all. Version surfaces and JSON validity: Targeted pytest: PR checks on 1a8adf6, via gh pr checks 609 --repo cdeust/Cortex, all pass or intentionally skip (Docker Build x2 and Tag a release skip as expected on a non-tag-triggering PR event; everything else, including Craftsmanship Gate, Type Check, Lint, all four Python version test jobs, SQLite backend, Windows, Fuzz batches, and pins current vs latest releases, is pass). No pending or failing checks remain. The round-1 blocking finding is closed, nothing else regressed, and every measurement round 1 held is unchanged at 1a8adf6. Merge is clear. |
Promotes the Unreleased section to 4.23.0 and moves the seventeen version surfaces (pyproject, manifest, plugin and marketplace manifests, Codex and Gemini manifests, package.json and lock, server.json, badge, uv.lock). Since 4.22.0: session hooks inject only the session's project and global memories (#604), one saved backend for Claude Code and Codex (#600), prediction records and calibration (#597), the wiki draft tools over MCP (#579), procedural skills on SQLite (#596) with a transcript-fed tool sequence (#591), the console entry point and Codex event adapter for the lifecycle hooks (#605, #608), the doctor worktree check (#602), and the launcher fixes (#573, #575, #582). The marketplace pin for zetetic-team-subagents moves to its released 2.42.0. Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_01Cz7A6goxVzfKL9NT7k4zDH
…vertises The review of this cut found version 2.42.0 beside ref v2.41.0 and the 2.41.0 sha; installs would have checked out 77b4f785 under a 2.42.0 label. ref and sha now name the v2.42.0 tag commit 96454db1, as the earlier deliveries (#553, #554) moved all three fields together. Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_01Cz7A6goxVzfKL9NT7k4zDH
The ADR-1082 entry cited a consolidate_launch module the craftsmanship gate refused; the shipped design relaunches session_lifecycle with --consolidate <mode>, as ADR-1082 and the code say. Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_01Cz7A6goxVzfKL9NT7k4zDH
1a8adf6 to
c293fd4
Compare
|
ZETETIC-REVIEW: COMMENT This supersedes the round-2 comment at 1a8adf6 (APPROVE, #609 (comment)). The branch was rebased onto main at 70343b6 to carry PR #610 (Codex rollout reader, SessionEnd detach, ADR-1081/1082) into 4.23.0, and the tip is now three commits: the cut, the round-2 pin fix, and a changelog wording fix naming the shipped consolidation launcher. Measured on c293fd4, matching the PR's current head. Merge base: Matches the stated rebase point. Diff against origin/main, 13 files, same release set round 1 and round 2 measured: CHANGELOG carries #610's entries under 4.23.0, and Unreleased is empty: Nothing sits between the two headers. The 4.23.0 section leads with the ADR-1081 entry for the Codex rollout reader (#610), confirming the rebase pulled the section in rather than leaving a stub. returns nothing (exit 1). The changelog wording fix in the third commit did what it says: it names the launcher that shipped, not the unshipped Version and pin checks: Only the one allowed failure, plus the matching NOTICE. The zetetic-team-subagents pin round 1 and round 2 dealt with does not appear. Both ADR-1081 and ADR-1082 mirror pages are present at wiki/adr/cortex/1081-read-a-codex-rollout-the-same-way-a-claude-transcript-is-read.md and wiki/adr/cortex/1082-sessionend-writes-its-session-log-before-it-detaches-consolidation.md, and the wiki checker confirms they match main. Where this stops short of a clean APPROVE: Everything measurable right now, the rebase point, the file set, the changelog content, the pin state, the doc claims, the wiki mirrors, and the script test suite, is clean and matches or extends what round 1 and round 2 already established. The remaining eight CI jobs need to finish and report green before this clears to merge; nothing so far suggests they won't. |
|
ZETETIC-REVIEW: APPROVE This upgrades the round-3 comment at c293fd4 (COMMENT, #609 (comment)), which held every measurement clean but stopped short of approval because eight CI jobs were still pending. Same sha, all other measurements from that comment stand unchanged (merge base 70343b6, the 13-file release set, empty Unreleased with the 4.23.0 section carrying #610's ADR-1081 entry, no consolidate_launch reference, 17 version surfaces agreeing, the single allowed PIN_VERSION_UNPUBLISHED failure for hypermnesia-mcp 4.23.0, clean doc claims, matching wiki mirrors for ADR-1081 and ADR-1082, and 1006 passed / 359 subtests passed in tests_py/scripts). gh pr checks 609 on c293fd4 is now terminal: 25 pass, 4 skipping (the two Docker image builds, the tag job, and the scheduled fuzz batch, none of which run on a PR event), 0 fail, 0 pending. The eight jobs that were still in flight when round 3 posted, the four Python version test legs, the SQLite and Windows SQLite test legs, and the two Fuzz PR-batch legs, all completed and passed. Round 1's blocking finding is fixed, round 2's rebase-and-wording-fix measurements hold, and CI is now fully green on the head this PR will merge at. Merge is clear. |
What
Promotes
## [Unreleased]to## [4.23.0] - 2026-09-17and moves every version surface to 4.23.0 (scripts/check_version_surfaces.py: 17 sites agree with pyproject.toml). Same file set as the 4.22.0 cut (#570), plus the README sentences that said "from the release after 4.22.0" now say "since 4.23.0", and the marketplace pin for zetetic-team-subagents moves 2.41.0 to 2.42.0 (released today).Since 4.22.0, on main: #604 (project-scoped injection), #600 (one saved backend for Claude Code and Codex), #597 (predictions and calibration), #579, #596, #591, #594, #582, #583, #585, #587, #589, #578, #573, #575, #602 (doctor worktree check), #605 and #608 (lifecycle hooks runnable from the wheel under Codex).
After merge: tag
v4.23.0on the merge commit;release.ymlbuilds and attests the distributions, publishes to PyPI and to the MCP registry.check_marketplace_pins.pyreportsPIN_VERSION_UNPUBLISHEDuntil that tag exists, as it did for #570.Measured on this branch
🤖 Generated with Claude Code
https://claude.ai/code/session_01Cz7A6goxVzfKL9NT7k4zDH