fix(code-tidying): backtick audit-comment-residue shape examples (0.14.3)

[kyle-sexton]

Closes #3191

Summary

Three rows of audit-comment-residue's shape-definition table wrote their example phrases in plain double quotes. audit-noise strips inline-code spans before matching, so those rows flagged against their own definitions on every sweep. The sibling skill already writes the same examples in backticks and does not self-match.

Fix

Wrapped the quoted example phrases in the plan-reference, conversational-antecedent, and ticket-pr-residue rows in backticks. No detector behavior change. Plugin version 0.14.2 → 0.14.3 with a changelog entry.

Verification

• audit-noise detect.sh against origin/main's copy of the skill: T1=2 T2=1 (the three named rows: plan-reference L38, conversational-antecedent L39, ticket-pr-residue L40).
• Same detector against this branch's file: Summary total: files=1 T1=0 T2=0 T3=0.
• scripts/check-changelog-parity.sh --check / --check-bump origin/main / --check-order / --check-preserved origin/main: PASS.
• scripts/check-changed-skills.sh origin/main: PASS (0 errors, 1 pre-existing Gotchas warning).
• scripts/check-skill-count-claims.sh --check: PASS.
• scripts/check-skill-leaf-names.sh --check: PASS.
• scripts/check-cross-plugin-source-drift.sh --check: PASS.
• markdownlint-cli2 on both changed markdown files: 0 issues.
• claude plugin validate plugins/code-tidying: Validation passed.

Related

• Refs feat(docs-hygiene): audit-noise gains three residue shapes (0.20.0) #3162 — where the self-matches were found and adjudicated; that PR was fenced to plugins/docs-hygiene/.
• plugins/docs-hygiene/skills/audit-noise/SKILL.md — the backticked sibling rows this matches.

[claude]

Claude finished @kyle-sexton's task in 31s —— View job

---

Security review — skip gate hit

• Gather context (diff vs origin/main, changed files)
• Apply skip gate

Result: no security review performed — skip gate condition 2 (no security-relevant surface).

All three changed files are documentation/metadata only:

• plugins/code-tidying/.claude-plugin/plugin.json — version bump 0.14.2 → 0.14.3.
• plugins/code-tidying/CHANGELOG.md — new changelog entry.
• plugins/code-tidying/skills/audit-comment-residue/SKILL.md — wraps three example phrases in a shape-definition table in backticks so they no longer self-match the audit-noise detector's plain-quote pattern. No detector logic, script, or workflow file is touched.

No code execution paths, credential/token handling, injection surfaces, access-control logic, or GitHub Actions configuration are affected by this diff. Per this lane's skip gate, no further review or findings are reported.

[github-actions]

Last security-reviewed head: 735a2d098726e845b5ca0e4dc8c99a79b456d1c0. On the next push, the relevance gate compares only the commits since this SHA; delete this comment to force a full re-review.

[claude]

Claude finished @kyle-sexton's task in 1m 25s —— View job

---

Code review complete

• Gather context (PR diff, changed files)
• Apply skip gate
• Review changed files against CLAUDE.md / REVIEW.md criteria
• Post findings

No findings. This is a clean, correctly-scoped fix.

What I checked

• Confirmed the mechanism the PR describes: plugins/docs-hygiene/skills/audit-noise/scripts/detect.sh skips YAML frontmatter entirely (in_frontmatter handling) and audit_noise_strip_inline_code (in lib/noise-shapes.sh) strips backticked spans before shape matching — so wrapping the three example phrases in backticks does make them invisible to the detector, exactly as claimed.
• Verified the fix touches only the three rows (plan-reference, conversational-antecedent, ticket-pr-residue) that actually have matching shape patterns in audit-noise's pattern set; the untouched history-narration row has no equivalent shape there, which is consistent with why it wasn't self-matching and didn't need the same treatment.
• Confirmed the sibling plugins/docs-hygiene/skills/audit-noise/SKILL.md already backticks its own example phrases in the corresponding shape rows, so this PR brings audit-comment-residue to the same established convention rather than inventing a new one.
• Checked the untouched frontmatter description line (plugins/code-tidying/skills/audit-comment-residue/SKILL.md:2), which still contains the same unbackticked example phrases — not a concern, since frontmatter is structurally exempt from audit-noise scanning (confirmed above), so it can't self-match regardless of quoting style.
• plugin.json version bump (0.14.2 → 0.14.3) and CHANGELOG.md entry are consistent with the plugin's existing versioning/changelog conventions and correctly describe a documentation-only change with no detector-behavior impact.
• Diff is minimal (3 rows re-quoted, changelog entry, version bump) — no scope creep, no unrelated changes.

I was unable to execute detect.sh directly in this sandbox (Bash execution for that script was not approved), so the exact T1/T2 counts in the PR's Verification section are author-claimed and not independently re-run by me — but the static trace through detect.sh/lib/noise-shapes.sh above corroborates the claimed mechanism and result.

[github-actions]

Claude has reviewed this PR 2 times. The lane skips further automatic reviews after 5; deleting this comment resets the count.

[claude]

Claude finished @kyle-sexton's task in 48s —— View job

---

Security review — skip gate hit

• Gather context (diff vs origin/main, changed files)
• Apply skip gate

Result: no security review performed — skip gate condition 3 (this head already has a successful security review that still applies).

Diffed origin/main...HEAD at the new head SHA 735a2d098726e845b5ca0e4dc8c99a79b456d1c0 against the diff already reviewed at the prior head 8ff20fb: byte-identical across all three changed files —

• plugins/code-tidying/.claude-plugin/plugin.json — same version bump 0.14.2 → 0.14.3.
• plugins/code-tidying/CHANGELOG.md — same changelog entry.
• plugins/code-tidying/skills/audit-comment-residue/SKILL.md — same three rows re-quoted into backticks.

The SHA change with no content change indicates a rebase/resync onto current origin/main rather than a new code change. No code execution paths, credential/token handling, injection surfaces, access-control logic, or GitHub Actions configuration are touched by this diff (unchanged from the prior review). Per this lane's skip gate, no further review or findings are reported.

[claude]

Claude finished @kyle-sexton's task in 26s —— View job

---

Code review — skip gate hit

• Gather context (pre-fetched PR data)
• Apply skip gate
• Post result

Result: no review performed — skip gate condition 1 (PR is closed or not open).

The pre-fetched context for this trigger shows PR #3215 as MERGED. Per /review:code-review's skip gate, review stops before deep analysis when the PR is closed or not open, and nothing further is posted.

For reference, this PR already received two prior automated reviews while open (security-review skip gate hit — docs/metadata-only diff; code-review — no findings, clean and correctly-scoped fix), both visible above in this thread.