Slice 1's event is held whole by its test and its mutants (#198's ruling 6, second of two)

Merged
DO
6d ago
+240-1

User description

This holds slice 1's event whole: the test of release/fixtures/event.slice-1.json and its mutants, as ruling 6 of "The ruleset switch: main requires the machine user's approval of the most recent commit, and slice 1's event is recorded" (#198) asks. It is the second of two; the fixture itself merged alone as 2054a9d2 in "Slice 1's event, the fixture alone: release/fixtures/event.slice-1.json (#198's ruling 6, first of two)" (#733). Part of #198; it does not close it.

Model-Lens inputs touched: none. The change is tests/test_release.py and tools/mutants.py.

Size. 25,659 characters and 284 lines by gh pr diff; 28,473 characters as the Kernel counts a diff, with its lines numbered, against a budget of 60,000. This pull request first held the fixture too and was held too_large at 67,308 by that count, so it was closed, the fixture went first, and it is reopened with the test alone.

The event

"Rulesets: main counts the machine user's approval, as the switch of 2026-11-11 left it (#198, record 0191)" (#730), on 2026-11-11:

WhenWhatBy
02:18:03Zmain last changed: it counts one approvalThe switch
03:13:04ZThe pull request openeddon-petry
03:26:49ZPass p_KQ88EB3ZFFWPE132C4EK: verifying, awaiting_checksfactory/v0.1.6
03:38:29ZPass p_NANTVRFXGVRZ0YMK958X: verifying, awaiting_review, admits the review Runfactory/v0.1.6
03:38:33Z to 03:40:46ZReview Run 42954100593, success; the Role's review of the head at 03:40:33Zcombsmith-sf-pr-review
03:42:19ZPass p_3RBP88Z3BEKDSTMRXSDX: converged, approval_due, acts issue_approvalfactory/v0.1.6
03:42:56ZReview 5853546913: APPROVED at 8fc54927donpetry-bot
03:43:04ZAdded to the merge queuedon-petry
03:45:08ZPass p_11GYAWVG130RGYV4Y1X1: queued, in_merge_queuefactory/v0.1.6
03:57:41ZMerged as d9eeb542The queue

factory/v0.1.6 was cut at 23:45Z on 2026-11-10, before the pull request's first commit at 02:24Z.

What the fixture holds, now on main

  • Twenty-two reads made afterwards through the Maintainer's login, each trimmed to the fields kept and passed through tools/redact_trial.py outside the repo; only the clean output was read. They are the ruleset, the repository, the pull request, its commits and reviews, the queue and merge events, next and the Release's tag and commit, each step's poller run and its artifacts, the review Run and the approving act with theirs, and the status on the item.
  • steps: the four passes whose Decision on the merged head differed from the pass before's, each with the head it read and its pass record's artifact id and digest.
  • event: the summary the test holds to the reads, with each Run's own outcome/1.
  • No pass record, applied, result or verdict file is kept. removed says what was taken.

Where it departs from ruling 6, for a ruling

  • Which pull request. Ruling 6 says "the first one after the switch". Two merged on the approval before this one: "P-58's deleted revision: GitHub still lists it, with its time, and a deletion moves nothing (#549)" (#726) and "The Pause's port says why each scope it could not read was unreadable (#688's dry runs 5 to 7, first of two)" (#729). #730 is taken because it is the switch's own record, the pull request rulings 1 and 4 name. The fixture's read says so. "First" is no read's to show, and ruling 6's wording is the Factory Builder's to amend.
  • Who enqueued. The Maintainer's login, as ruling 4 says. The Factory's queue step has not enqueued a pull request, and the test holds the enqueuer to the Maintainer.
  • The fixture's form. A plain fixture with top-level exchanges, as refs.read.json is, not release-fixture/1 with a kind: that schema as built holds only runs of release.yml.
  • The status at each step. The one check run is edited in place, so it is read back once, after the merge, and shows the last step. The earlier steps rest on their records' digests, which GitHub keeps until 2027-02-09.
  • next at each pass. It is read once, afterwards; where it stood at each pass is that pass's own word.

Checks it adds

tests/test_release.py::test_the_slice_1_event_is_whole holds, against the fixture's own reads and the configuration:

  • the ruleset counts one approval by all five fields, as config/rulesets.json records, covers the default branch, which is the pull request's base, and last changed before the pull request opened;
  • the author is the Maintainer; the one approval is the approver's, a user's, of the head, of the last push, and the last review made, with no request for changes;
  • the Maintainer's login enqueued it after the approval, and the merge is the pull request's own, into main;
  • every step and both Runs name one Release, the one next and its tag name, cut before the first commit, by the cut's own fixture;
  • each step is a distinct pass in strict order, of the head that merged, its poller run on main, its record the artifact GitHub lists;
  • the review Run and the approving act carry the item, the Intent, the Role, the one content and the Run's id in their names, ran on main and succeeded; the Role's own review is of the head, within its Run;
  • the status on the item is the controller's, and is word for word what the effects port writes for the last step's Decision.

Thirty-two mutants of the fixture are killed by it. release/AGENTS.md gains no row: it is at its 40-line limit.

No record is needed: no rule changes.

Checks

CheckResult
make checksOK
tests.test_releaseOK, 29 tests, on the head with main merged in
Mutants of the fixture, run alone on an export32 of 32 killed, at fe634c3
gitleaks on the fixtureNo leaks
Full make verifyNot run locally, by the Maintainer's standing change of 2026-11-09; CI's verify is the answer
ReviewerHeadUnread partsBlockingMajorMinorInfo
agy (Carbon)a27c4danone0230
musea27c4danone0310
astraa7da8b5none0200
astra793a27enone0100
astra11e82afnone0200
claude-reviewa27c4danone0142
claude-securitya27c4danone0081

The readers read fb3d5ce and before; the head is fe634c3, which adds only the merge of main, where the same fixture now is. Every finding was a claim the test did not hold, or a read the fixture did not keep; none found the fixture wrong, and the Claude review checked the four steps against the pass records live. Astra's last two majors, at 11e82af, are of that kind and are fixed at the head; by the lane's rule a gap in a test needs no read again, so no reader read again after them.

FindingWhatDisposition
muse #1, security #5, #6, astra@a7da8b5 #1, astra@793a27e #1A read was trusted by its path; the repo, each run's id, each tag's ref, the query's variables, the ruleset's branch and the default branch were unreadfixed: each is held; the fixture keeps the ruleset's conditions and the repository's read
muse #2, claude-review #1, security #4Nothing held the event to after the switchfixed: the ruleset's own read, last changed before the pull request opened
muse #2, security #9, agy #1, astra@11e82af #2The approval was not held to the last push, nor a step to the head that mergedfixed: the last commit is the head; the approval is the last review; each step keeps the head its pass read
muse #3, claude-review #2, security #1, #2, agy #3, #4The review Run was not tied to the approved content, nor either Run to mainfixed: each name is compared whole; both ran on main; the Role's own review is of the head
muse #4, security #3, #7, agy #5, astra@a7da8b5 #2The status was held by its title, pass and Release alonefixed: its App, and its whole output as the effects port writes it
claude-review #5Each Run's own outcome was droppedfixed: kept with its artifact and digest, and held to the event's Release and no fault
claude-review #6, security #8, agy #1FR-50 compared the commit's date, not the cut'sfixed: the cut's own fixture is read
agy #2Steps could repeat a passfixed: distinct pass, run, artifact and digest, in strict order
astra@11e82af #1The ruleset's fields were compared by whichever the fixture keptfixed: all five are named
claude-review #3Three pass ids are held by nothingno change to the test: removed says each step's id, State and codes are transcribed and rest on its record's digest
claude-review #4By ruling 6's words the event is #726the fixture's read, the test and this description say so; the ruling is the Factory Builder's to amend
claude-review #7release/AGENTS.md has no row for the testno change: the file is at its 40-line limit

🤖 Generated with Claude Code


CodeAnt-AI Description

Verify the recorded approval and merge history for slice 1

What Changed

  • Add checks that the pull request was approved by the configured approver on its latest commit, after the approval requirement took effect.
  • Verify that the merge, release, review, and recorded workflow steps all match the same event and occur in the expected order.
  • Catch altered event records, including mismatched approvals, release details, workflow results, and status checks.

Impact

✅ Catches approvals on the wrong commit
✅ Detects mismatched release and merge records
✅ Flags incomplete event histories

💡 Usage Guide

Checking Your Pull Request

Every time you make a pull request, our system automatically looks through it. We check for security issues, mistakes in how you're setting up your infrastructure, and common code problems. We do this to make sure your changes are solid and won't cause any trouble later.

Talking to CodeAnt AI

Got a question or need a hand with something in your pull request? You can easily get in touch with CodeAnt AI right here. Just type the following in a comment on your pull request, and replace "Your question here" with whatever you want to ask:

@codeant-ai ask: Your question here

This lets you have a chat with CodeAnt AI about your pull request, making it easier to understand and improve your code.

Example

@codeant-ai ask: Can you suggest a safer alternative to storing this secret?

Preserve Org Learnings with CodeAnt

You can record team preferences so CodeAnt AI applies them in future reviews. Reply directly to the specific CodeAnt AI suggestion (in the same thread) and replace "Your feedback here" with your input:

@codeant-ai: Your feedback here

This helps CodeAnt AI learn and adapt to your team's coding style and standards.

Example

@codeant-ai: Do not flag unused imports.

Retrigger review

Ask CodeAnt AI to review the PR again, by typing:

@codeant-ai: review

Check Your Repository Health

To analyze the health of your code repository, visit our dashboard at https://app.codeant.ai. This tool helps you identify potential issues and areas for improvement in your codebase, ensuring your repository maintains high standards of code health.