Repository navigation
Trials P-30 and P-63: the release identity and the eval job, recorded from their runs (#189, #309) - #667
Trials P-30 and P-63: the release identity and the eval job, recorded from their runs (#189, #309)#667don-petry wants to merge 6 commits into
Conversation
… the tag ruleset (#189) trials/P-30.json, from release runs 42348605675 (the cut of factory/v0.1.0 and the move of factory/v0-next) and 42348650504 (the re-cut, refused), read again through the Maintainer's login. The Deployment the cut recorded names combsmith-sf-release[bot] as its creator. A live read of tag ruleset 25183656's bypass list names the admin role and App 5502359 alone, and stands in for each other identity's refusal; no other identity's refusal was exercised, as the pass condition says. Decision 0114 is why the App and not the poller's job token. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…t by hashes, and its environment admits main alone (#309, record 0171) trials/P-63.json, from eval run 42347416962, the gate's run for v0.1.0: its run, jobs and artifact, the report's commit, set and each case by id and sha256 alone, and a live read of factory-eval's deployment branches (main, no admin bypass), which stands in for a dispatch from another branch refused. A new section-6 row asks it; TRIALS_OF_THE_FACTORY names it beside P-30, with a test and a mutant; record 0171 amends 0113's "P-30 alone". The record number is provisional until the enqueue. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…nd record 0171 says the conditions came after the runs From the reviews of 836f22d: the claim that the admin role is the Maintainer's login alone is now an exchange of its own (don-petry alone); P-30's row in section 6 gives its result as the other rows do; record 0171 states that each pass condition was written after the Release's runs, setting aside the skill's step 1 for these two fixtures alone; and two notes name P-63 and decision 0113 where they should. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…tays provisional until the enqueue Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Important Review skippedAuto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the
 |
Codex Review Summary: ❌ FailedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
…y Builder gave 0173 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Closes the fixture boxes of "Trial P-30: one identity cuts a Release and moves a channel tag under the tag ruleset" (#189) and "Workflows: eval.yml runs the review Intent's held-out cases and writes eval-report/1" (#309), under the slice-1 epic (#182). By the Factory Builder's rulings of 2026-11-09, pre-approved by the Maintainer on 2026-11-08. Both issues are closed by hand once this merges, with a comment each.
What
trials/P-30.jsonfactory/v0.1.0, movedfactory/v0-next) and 42348650504 (the re-cut, refused), read again through the Maintainer's login. The cut's Deployment namescombsmith-sf-release[bot]as its creator. Live reads of tag rulesets 25183656 (bypass: the admin role and App 5502359 alone) and 25996980 (no bypass), and of the repo's admins (don-petry alone). Verdictpass, with the residual stated: no other identity's refusal was exercised. Decision 0114 is why the App and not the job tokentrials/P-63.jsonmainat 499835d: run, jobs (fetch and upload succeeded), artifacteval-report(unexpired to 2027-02-05). The archive's sha256 is GitHub's digest. Of the report: commit, Release, set hash and count, totals and the one-word bars, and each case by id and sha256 alone (they equalevals/pr-review/cases.jsonat the commit). Each case's verdict and reasons are left out, since a reason tells a case's kind. Live read offactory-eval: one branch policy,main, andcan_admins_bypassfalse. Residual stated: no dispatch from another branch was exercised. No secret, and no list of secrets, was readtools/check_trials.pyTRIALS_OF_THE_FACTORY = {"P-30", "P-63"}tests/test_tools.py,tools/mutants.pytest_the_eval_jobs_trial_records_the_factorys_own_repo, and the mutant "the eval job's trial is no trial of the Factory"trials/AGENTS.md, the trial skill,schemas/trial.schema.jsonNeither fixture grounds a rule of
sim/grounding.py: none names P-30 or P-63.Also done on GitHub, outside this PR: "Trials P-2, P-5, P-23 and P-25" (#185) closed, its grounding box amended; its follow-on filed, "Trial: the merge queue merges only after a pass has read the pull request" (#659).
Verification
Export verify of the head: pass at
9e58a78a(every check and every test). Mutants on the new rule, run alone on an export: 3 of 3 killed ("the eval job's trial is no trial of the Factory", "any trial may record the Factory's own repo", "a trial of the Factory may record the canary").python3 -m tools.check_trials: ok.Reviews, each of
836f22d; the findings were fixed in5ff92161(rebased on #657, then main merged and the record renumbered to 0173 in9e58a78a):836f22d836f22d836f22d836f22ddocs/architecture/software-factory-design.md:2223trials/P-30.json:5tools/mutants.py:2659trials/AGENTS.md:14trials/P-63.json:5trials/P-30.json:5🤖 Generated with Claude Code