[kyle-sexton]

Refs: #12689

Summary

Replaces #12783, the work-resolve-issue lane's pull request for #12689, and carries its diff
unchanged plus one new commit.

The merge queue refused #12783's head 4a0b6a3a with verify:earlier-fail: pr-verify FAILed
an earlier head, 4ec5632c, and that FAIL carries forward to every head with the same change
digest. 4a0b6a3a was an empty refire commit, so its digest matched. The lane could not clear it:
both address-feedback rounds pushed nothing, because the fail-note threads were already resolved
and left no open thread to act on. This pull request answers the fail-note findings with a real
change, so its digest differs from the FAILed one.

Fix

One commit on top of #12783's diff, answering each pr-verify fail-note on 4ec5632c:

• report.md row r080 (the finding that failed the head). The Routes item relayed
  /skill-doctor and /doctor coverage to SKILL.md "Scope boundary (route out)", which the issue
  says holds the same restated text, with no as-of date or trigger. It now points at both commands'
  rows in commands § All commands, as of
  2026-11-03, rechecked when either row is renamed, removed, or changes what it covers
  (plugins/context-budget/skills/audit/reference/report.md:58-62). I read the page this session:
  the /doctor and /skill-doctor rows are both under All commands.
• SKILL.md:78, PR body overstated the native-explain-usage.md change. That file is now in
  the diff. Its absence row named only the commands page and changelog, as of 2026-10-11, while the
  skill body says commands, skills and the docs index as of 2026-11-03. The row now lists every page
  read (commands § All commands, skills § Bundled skills, llms.txt, the changelog) as of
  2026-11-03, with a matching trigger
  (plugins/context-budget/skills/audit/reference/native-explain-usage.md:11). All four pages,
  fetched this session, have no explain-usage mention. claude-review flagged the same mismatch as a
  minor note on docs: replace restated upstream facts with pointers in work-items, context-budget #12783.
• check-stamps.sh not run. It ran here and found nothing (see Verification).
• README.md:197 row r069, relay through engine.md. No change. engine.md "Comparability rules"
  holds the stamped /context pointer and records the subtraction as the engine's own observation,
  because the /context row does not itemize the bucket. The verifier called this acceptable.
• setup SKILL.md:425 row r481, providers.md claim unverified. No change needed:
  plugins/work-items/skills/setup/reference/providers.md:61-63 carries the Atlassian Issue
  resolutions pointer with an as-of date and recheck trigger.
• Adapter README.md:280 --remove-assignee "@me" line. No change: the verifier noted it for
  awareness only, and it matches row r479.

Row r478 (the attend-queue rate-limit floor) stays open, as on #12783.

Verification

• scripts/check-affected.sh --base origin/main --no-fetch: 13 passed, 2 skipped (no shell, no
  Python) over 15 changed files.
• plugins/attribution/skills/audit/scripts/check-stamps.sh --trigger-less on the 13 touched
  markdown files under plugins/: 62 stamps parsed, 0 findings, 0 declined.
• markdownlint-cli2 on the two edited files: 0 issues.
• The pre-push hooks (check-affected, reference-verify, pre-push content scan) passed.

The AI review lanes are down repo-wide until a token swap, so the review checks and ci-status
are expected to fail on this head.

Related

• Replaces docs: replace restated upstream facts with pointers in work-items, context-budget #12783
• docs: replace restated upstream facts with pointers in work-items, context-budget #12689

🤖 Generated with Claude Code

https://claude.ai/code/session_01Vcupom3BKF6QuC5U3sRJPj

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-03T10:05:11.171845Z	7e82a79	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[github-actions]

claude-security-review reviewed a5d19433ec8942dfebcaea1020924edf01a268c1: read 2 of 2 changed file(s); 0 finding(s), 0 as inline comments.

Security review: PR #12810, incremental (7e82a796 → a5d19433)

No security issues found.

Scope. I read every hunk in both files on the incremental list. Both are markdown, and neither matches a TRUSTED PATHS glob. The push changes no code, workflow, action, plugin manifest, userConfig key or hook.

Removed or reworded instructions. I could not read the protected-class list in docs/contracts/instruction-exception-register/README.md this run (no shell), so I judged each change by what would break without it. Neither change removes or weakens a rule:

• plugins/context-budget/skills/audit/SKILL.md: the boundary bullet for the native explain-usage skill drops the as-of date and recheck note it repeated. It now just points at reference/native-explain-usage.md, which already holds that record. Nothing the skill forbids or gates changes.
• plugins/work-items/reference/permission-preflight.md: the text now says why candidate 2 (core.worktree) does not resolve for --separate-git-dir repositories. Git leaves that key unset in all three ways such a repository can be created, observed on git 2.53.0, with a recheck trigger. This only explains an existing failure case. The verification predicate, and the warning that git's worktree list misreports the main worktree here, are unchanged.

PR context. The pull request text contains no instructions aimed at this review.

Prior findings. The one earlier post is this lane's clean review of 7e82a796. It had no findings, so there is nothing to repeat.

critical-findings: 0

[github-actions]

claude-review: 2 inline finding(s) on 7e82a796683edb7cbf458c435c195c058b99dcaa; the summary is in this lane's pull request comment.

claude-lane: claude-review

[github-actions]

claude-review reviewed a5d19433ec8942dfebcaea1020924edf01a268c1: read 2 of 2 changed file(s); 0 finding(s), 0 as inline comments.

Code review: PR #12810, incremental review (7e82a79 → a5d1943)

Scope: I read every hunk of both files in the incremental change (.claude-lane/incremental.diff): plugins/context-budget/skills/audit/SKILL.md and plugins/work-items/reference/permission-preflight.md. Both changes are Markdown only. Neither file matches a trusted-path glob. The PR closes no issue, so there are no acceptance criteria to check. This review is round 2 of a cap of 3, so the cap does not apply. The settled threads and prior findings contain no directives aimed at this lane.

Prior findings, both fixed by this push:

1. plugins/context-budget/skills/audit/SKILL.md:78-80: the bullet no longer carries its own as-of date or recheck trigger. It now points at reference/native-explain-usage.md, whose row 11 holds the single date and the full trigger. That removes the second stamp whose trigger missed some sources. The owner has also resolved this thread, and nothing has changed since the resolution.
2. plugins/work-items/reference/permission-preflight.md:221-225: the repo's own observation is back. It names the three creation paths, the git version (2.53.0), an as-of date and a recheck trigger, and it now states outright that "candidate 2 does not resolve". The "therefore … resolves to <path>" conclusion has its premise again.

New findings: none.

Verification: I ran nothing. I did not re-run the git observation the author records (git 2.53.0, 2026-11-03), and I did not fetch any upstream page.