[Dong1017]

Governing Issue and delivery slice

Refs #273

• Accepted scope / delegation: R1 as recorded in the owner's decision: a pr-references job in ci-ready.yml on the ready lane's existing events, not a required check, delivered by one PR without a separate task Issue.
• What this PR delivers: a checker for closing syntax in PR body text, its tests, the ready-lane job, and short updates to working rule 6, the contributing guide and the PR template.
• What it does not deliver: checks of commit messages or Development sidebar links, an edited trigger, a required status, or R2–R6 of [Investigation] Agent-led workflow gaps: closure, decision visibility, review provenance and focus #273.

Knowledge authorship declaration

Not applicable: this PR changes no path protected by docs/knowledge-model-policy.md.

Changed artifacts and evidence

Requirement	Actual change	Check / evidence and result	Remaining gap
Closing syntax inside PR body text fails, including negated sentences	tools/check_pr_references.py	tests/test_pr_references.py uses the eight negated sentences from #273 F1; all are reported. Local run passed	none
A line holding only a keyword and one reference passes	same	same test file, including cross-repository references, issue URLs and CRLF bodies	none
The PR template itself passes	.github/PULL_REQUEST_TEMPLATE.md	test_pull_request_template_passes	none
Runs in the ready lane and does not block merge	.github/workflows/ci-ready.yml, job pr-references: read-only token, body passed through env, printed lines start with line so PR text cannot form a workflow command	this PR's ready run	not a required check by decision
The rule is documented	docs/agent-collaboration.md rule 6, docs/contributing.md, PR template comment	review	none

Why the existing ready-lane events suffice: for each of the eight closures in #273 F1, the closing sentence was already in the PR body at the last ready-lane event before merge. This was checked on 2026-10-31 by comparing each PR's body edit history (GraphQL userContentEdits) with its ci-ready run times and timeline events. A body edit after the last push is checked only at the next push.

Checks and dependencies

• Checks actually run (result and tested scope): on Windows 11 with Python 3.13.5, on a tree identical to this head: pytest tests/test_pr_references.py tests/test_qualify_kv_reclamation.py tests/test_qualify_kv_edit_actor.py tests/test_knowledge_model_policy.py gave 187 passed and 3 failed. With PYTHONUTF8=1, the two tests that read ci-ready.yml pass, and tests/test_pr_references.py passes. ruff check src tests tools passed.
• Failed, skipped, timed-out or not-run checks (if applicable): test_retained_actual_outputs_have_labelled_path_redactions fails on Windows because it expects POSIX paths; this PR changes neither that test nor its code. Without PYTHONUTF8=1, the two ci-ready.yml readers fail on a GBK locale because the file's first line already contains an em dash. The full suite was not run locally; the ready lane runs it on Linux.
• Dependencies / shared interfaces / integration assumptions (if applicable): the KV live policy pins one specific ci-ready run ID, so an added job does not affect it. No existing job, trigger or concurrency setting changes.

Remaining work and Issue closure

• Remaining Issue requirements / follow-up: [Investigation] Agent-led workflow gaps: closure, decision visibility, review provenance and focus #273 stays open until R6 is decided.
• Goal-acceptance owner or accepted closure delegation: Dong1017
• Goal ready for acceptance? Not claimed

🤖 Generated with Claude Code

[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 2026-11-01T02:31:15.003236Z	b86caeb	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.

[Dong1017]

Scoped LGTM for the accepted R1 slice at b86caeb1. All six current-head Ready jobs and the declaration check passed; no confirmed blocker remains in the inspected delta. This is a review conclusion, not formal approval or acceptance of the other #273 remedies.

• Inspected all six changed files and the workflow invocation. Sentence-embedded/negated directives are reported; explicit standalone lines remain allowed. Existing events and required-check settings are unchanged.
• Independently checked the published-head parser with 172 cases in an isolated, empty environment. Documented keyword/case/colon/reference forms, absent input and malicious-body handling passed. PR text remains an environment value rather than shell source; reported physical lines receive a fixed prefix.
• The completed native review reports no findings, and the current review/thread lists contain no outstanding opinions requiring disposition. The author's Windows locale/path failures are not presented as a local full-suite pass; both Linux full-check jobs now pass.
• Limits: this is an advisory syntax check, not closure authorization or full GitHub parser equivalence. Cross-line keyword/reference recognition is not covered; GitHub recognition of that form was not established, so it is not asserted as a verified blocker. Commit messages and sidebar links remain outside accepted scope.

Inspected revisions and evidence

Head: b86caeb1f898bca7468ee4bea6803f47a4433f8e; base: 218565698d850e955f85d0d394a5f0e5e3619812.
Accepted decision, Ready CI, declaration CI, native review.

This review used the actual PR implementation, not the separate unpublished-PR local candidate. Independent tests were scoped to the parser; full-suite evidence comes from hosted CI. No model, device or paid experiment was run for this review. No reviewer thread was resolved or repository setting changed.