chore(pr-reviews): land the PR #2537 review fixes that missed its merge

[josecelano]

Lands the three review fixes for PR #2537 that missed its merge. #2537 merged at its approved rebased head before these commits reached it. The maintainer's approval for this follow-up is #2537 (comment).

Related to #2537 and #2536.

Changes

• open-pull-request skill: run the PR-branch commit-id check before every push that changes an issue spec, evidence file, or review audit, not only after a rebase (finding F18). For the record (finding F17): docs(issues): [#2536] add specification for hiding the statistics read lock #2537 also changed this skill, in "docs(skills): check every changed file for PR-branch commit ids". The check for PR-branch commit ids now scans every changed text file, keeps all-digit ids, and reports only ids that resolve to commits outside develop. docs(issues): [#2536] add specification for hiding the statistics read lock #2537's merge commit does not mention that change.
• project-words.txt: remove unreassigned, which is ordinary English. The docs(issues): [#2536] add specification for hiding the statistics read lock #2537 review audit ignores it file-locally instead, as the PR test(dev-tools): address PR #2339 post-merge review findings #2344 and PR fix(tracker-core): [#2406] scrape reports persisted downloads after a restart #2423 audits do (finding F19).
• docs/pr-reviews/pr-2537-review/PR-REVIEW.md: record the fourth review (F17-F19) as follow-up work, with the post-merge approval. The threads stay open until this PR merges, and the audit is closed out after it merges.

Pre-commit and pre-push checks passed through the installed git hooks.

[Copilot]

🟢 Approval recommended

The focused documentation changes match the approved follow-up scope, with no unresolved findings.

0 open findings

What changed in this PR

Lands the documentation and workflow fixes that missed PR #2537’s merge.

Changes:

• Requires the commit-ID check before documentation pushes, not only after rebases.
• Moves unreassigned from the shared dictionary to a file-local spelling exception.
• Records F17–F19 as approved follow-up work, pending merge.

File	Description
project-words.txt	Removes unreassigned.
docs/​pr-reviews/​pr-2537-review/​PR-REVIEW.md	Records follow-up approval, findings, and pending thread states.
.github/​skills/​dev/​git-workflow/​open-pull-request/​SKILL.md	Expands when the commit-ID check is required.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[da2ce7]

Reviewed at 51ca18affd7c0cfa198dc69917fc9fca0d82a342. Recomputed from the bytes at this head, against the merge base e6b46013b; develop is ee4eb8e9c and git merge-tree against it is clean.

F17 as landed. git log -1 --format=%B e6b46013b carries the original description, so the #2537 merge commit never mentions the skill change. develop history is never rewritten, so this PR's body, which goes into its own merge commit, is the only remaining durable place for that note. The body describes the earlier change accurately against the skill at e6b46013b^2: every changed text file is scanned, all-digit ids are kept, and only ids that resolve outside develop are reported. The first F17 reply ("Fixed by editing the PR body") was untrue once the merge commit existed, and the 06:57 follow-up reply corrects it.

F18 as landed. :78-81 make the check run "before every push that changes an issue spec, evidence file, or review audit, and before pushing after a rebase". That matches the artifacts the rule at :74-77 lists. The PR body and review replies cannot be scanned by a command over changed files, and :101 still says they are a manual read. The lines are at most 100 characters, and MD013 is off.

F19 as landed. unreassigned is gone, and unpushed and unrecognised are neighbours again. The directive sits two lines below the skill-link marker, as in the PR #2423 audit (:10, :12) and the PR #2344 audit (:11, :13). At the head, git grep finds the word only in those three audits, and each carries its own directive, so no file relied on the dictionary entry.

Audit update. The validator exits 0 offline on the #2537 record (19 rows, 8 log entries). It skips URL resolution references and checks only that each Reply URL is on its own thread, so the three FOLLOW_UP/OPEN rows pass. The record matches the capture. Review 5525009643 is APPROVED at 21:38:28Z, at the rebase head, with three Nit comments. The merge came at 06:44:22Z, the replies at 06:44:50-53Z, and the follow-up replies at 06:57:02-06Z. The approval is comment 6186327939, which the Ownership field (process-pr-review :217) now holds; it is retrospective and states its ~06:52Z approval time (:125-127). FOLLOW_UP and OPEN are defined in the record, and each detail's Follow-up PR URL names #2539, as :137-140 and :236-237 ask. The rebase-head legend entry is the merged head. "Fourth review" counts the four posted reviews. The 06:58 entry precedes its commit (06:58:15Z), and nothing above it changed. The body's audit bullet matches the diff. One category deviates (F1).

Process. No audit for this PR is due yet: the only review so far is an overview with no findings, which earns no row (:51-52). Related to #2537 and #2536 fits open-pull-request :156-157 for a follow-up that closes nothing, and the title rightly carries no issue tag (AGENTS.md:394). Each commit has one well-formed trailer line and no other body, author equals committer, and git diff --check is clean. The skill's id check, run verbatim in wt-head-2539-r1, prints nothing. The chore title type is F2.

Findings. F1 and F2 (Nit), inline. None is blocking.

Checked, no finding. develop's #2535 delta touches only the #2493 spec folder, none of these three files. The review bot's overview has no inline comments to number. A few new log lines run past 100 characters, which MD013 allows.

Checks

On the loop's compute hub at this head, merge base e6b46013b (receipt server-gates-krkavec-306): pre-commit profile gate exit 0 (80 s); linter markdown, linter cspell, linter lychee exit 0; frontmatter-validator --all: 19 errors / 4 warnings at this head and 19 / 4 at e6b46013b re-measured, none in the files this PR touches; validate-audit-record.py --pr-number 2537 exit 0 against the PR #2537 audit record at this head (19 rows, 8 log entries, 0 failures). GitHub checks at this head at posting time: "Docs Lint: completed/success". At posting, develop is ee4eb8e9c.

[da2ce7]

ACK 51ca18a — F17-F19 as landed, the PR #2537 audit's follow-up rows against the capture, and the clean merge against ee4eb8e were recomputed, and only non-blocking findings stand

[josecelano]

ACK 51ca18a

[josecelano]

Maintainer approval for the review findings that were unprocessed when this PR merged: @da2ce7's review #2539 (review) was submitted at 2026-10-15 07:15:52Z, and this PR merged at 07:21:44Z with its two inline findings still unprocessed.

• F1: change the PR docs(issues): [#2536] add specification for hiding the statistics read lock #2537 audit's F19 category from other to maintainability, matching the PR fix(tracker-core): [#2406] scrape reports persisted downloads after a restart #2423 audit's record of the same concern.
• F2: NO_ACTION. This PR's chore title is already the permanent merge subject on develop and cannot be changed. The PR docs(issues): [#2536] add specification for hiding the statistics read lock #2537 follow-up records the lesson instead.

Both land in one loop-closing PR from develop. That PR also closes out PR #2537's F17-F19, which this PR delivered. It creates docs/pr-reviews/pr-2539-review/PR-REVIEW.md for this PR's findings. The two threads stay open until it merges.