The fourteenth cut and the move of next: release/fixtures/cut.v0-1-13.json
This records the v0.1.13 cut as release/fixtures/cut.v0-1-13.json, in the same form as cut.v0-1-12.json. The Factory Builder asked for it on 2026-11-13. It closes no issue and carries no record.
Model-Lens inputs touched: none. Protected paths touched: none. The change is the fixture and tools/mutants.py, and goes by the Factory's route.
Size. 47,312 characters and 839 lines by a local diff against main; 54,563 characters as the Kernel counts a diff, with its lines numbered, against a budget of 60,000.
The fixture adds no line to tests/test_release.py. Since "A cut's fixture reads a page of next's Deployments, held to its order, and every cut after factory/v0.1.11 must (#792)" (#813) the test reads every cut's fixture from release/fixtures/ in version order, and the helper every later cut is read by is unchanged here.
What GitHub answered
- Release run 43494364478 (
release.yml, a dispatch onmain, success) cut factory/v0.1.13 at 4e58ba40, the merge of "A push past a Lens's review asks it again while its thread gates, and a thread its later review dropped is the Kernel's to resolve (#833, record 0212)" (#859). It movedfactory/v0-nextthere and recorded Deployment 7774706592, made at 19:40:26Z, whose status is success at 19:40:27Z. The fixture holds the run, its one job, and the job's log at the step. - Eval run 43491470154 (
eval.yml,workflow_dispatch, success) ran on 4e58ba40. It ended at 19:39:52Z, before the release run began at 19:40:09Z, and the eval workflow lists it as its only run on that commit. - Read after the cut (19:41:53Z):
- factory/v0.1.13 and
nextboth name 4e58ba40. - factory/v0.1.12 still names ed1a18ed, where its cut left it.
- The page of
next's Deployments holds four, newest first: v0.1.13's, v0.1.12's, v0.1.11's, then the move forward's of 2026-11-12. - The Deployment
nextmoved off is the thirteenth cut's own (7765100902). GitHub marked itinactiveat 19:40:28Z, after the new one was made.
- factory/v0.1.13 and
The reads went through the Maintainer's login and were trimmed to cut.v0-1-12.json's fields and headers, read for read. tools/redact_trial.py redacted them outside the repo, and only the clean output was read. It took out the names of steps[1] and steps[6], the token's mint and its post step, as for v0.1.12. removed says what was taken, and says one thing more than v0.1.12's does: the Link header GitHub answers on the page's read, which the trim to the older fixture's headers drops (the security pass's finding; see below).
Checks it adds
tests/test_release.py::test_the_next_cut_moved_next_onnow reads this fixture with the rest, and holds of v0.1.13 all that it holds of v0.1.12.- Forty-six mutants of the new fixture, each killed by that test. They are the thirteenth cut's forty-six, carried over value for value: each value of the old fixture changed for the one the new fixture holds at the same line. Three kinds were then set by hand, so that each breaks the rule its name gives and no other:
- the nine rows that turn on a time take the new fixture's own times;
- the five rows that change a run's or a Deployment's id to another take the new fixture's id and one more;
- the two rows of the page's third and fourth Deployments are renamed, since those are now the twelfth cut's and the move forward's.
release/AGENTS.md's cut row already names the test, and is unchanged.
Checks
| Check | Result |
|---|---|
make checks | OK, at a911946 |
tests.test_release | OK, 38 tests, at a911946 |
| The forty-six new mutants, each applied by hand | 46 of 46 killed, at a911946 |
| Mutants of the cuts' fixtures and of the rollback's, run alone on an export | 238 of 238 killed, at 4ad0b4a |
| gitleaks on the fixture | No leaks |
Full make verify | Not run locally, by the Maintainer's standing change of 2026-11-09; CI's verify is the answer |
Since 4ad0b4a one sentence of the fixture's removed changed, which no mutant's text holds, and the branch was rebased on main.
| Reviewer | Head | Unread parts | Blocking | Major | Minor | Info |
|---|---|---|---|---|---|---|
| agy (Carbon) | 4ad0b4a | not run (quota), owed | – | – | – | – |
| muse | 4ad0b4a | not run (the plan's usage limit), owed | – | – | – | – |
| claude-review | 4ad0b4a | none | 0 | 0 | 0 | 1 |
| claude-security | 4ad0b4a | none | 0 | 0 | 0 | 2 |
| astra | 4ad0b4a | none | 0 | 0 | 0 | 0 |
| Kimi | e4d4c51 | none | 0 | 1 | 0 | 0 |
Kimi: 1 finding, 1 taken. Both Claude readers asked GitHub again for the reads they could, and each kept field agreed; each applied all forty-six mutants one at a time and found each killed at the assertion its name gives.
| Finding | What | Disposition |
|---|---|---|
| Kimi #1 (major) | The mutant "the eval workflow lists the fourteenth cut's eval run as ended after the cut" kept the thirteenth cut's time, which is before this cut: it was killed, but not for the reason it names | fixed in 4ad0b4a: the new fixture's time, moved past the cut. The five ids carried as the old fixture's number and one were set to the new fixture's in the same pass |
| claude-review #1, claude-security #1 (info) | Five fields of a cut's fixture are read by no test, in v0.1.12's fixture as in this one: the release run's own status and attempt, the job's name, the environment of the cut's Deployment's status, and the times of the statuses under the Deployment next moved off | not here: it is the shared helper's, and a change of its own. Filed as "A cut's fixture has five fields no test reads, and drops the Link header of the read of next's Deployments" (#863) |
| claude-security #2 (info) | The read of next's Deployments drops GitHub's Link header, one a fixture keeps, and removed did not say so. Carried over from v0.1.12's fixture and the rollback's | this fixture's removed now says so. Whether a cut's fixture keeps and holds the header is in #863 |
🤖 Generated with Claude Code