[kyle-sexton]

Closes #6572

Summary

Rewrites the house-owned parts of two playbooks skills (boris, skill-authoring) to the skill criteria, wording only: 11 files, 0-2% each. Synced upstream content (boris/vendor/, skill-authoring/vendor/, the SKILL.md sync metadata) and quoted upstream wording (Boris's and Thariq's words) are untouched. No rule, threshold, number, example, identifier, cross-reference, frontmatter, code block or script changed.

Fix

Per-skill decision table:

Skill	Ships	Evidence
boris (SKILL.md + 7 reference files)	rewrite	2 labelers PASS, FALSE POSITIVE / RELOCATED rows only
fable-5	unchanged	doctrine Fable wrote for itself; /playbooks:update regenerates it only on a Fable model-version change, and #4346 reserves that for a Fable 5.1 session (an earlier commit here reworded it; reverted to main's text)
skill-authoring (SKILL.md + 2 reference files)	rewrite	2 labelers PASS, no SEMANTIC LOSS / AMBIGUITY row
repo-sweep, update	unchanged	they write state (commits, PRs, synced files); the lane rule gives a state-mutating skill full before/after measurement even for wording-only edits, so they are left as they are

Deliberately unchanged inside the three: skill-authoring/reference/skill-criteria.md (the house contract; six frozen eval cases cite its line ranges), rewrite-a-plugin.md (numbered runbook), and the boris chapters that are mostly upstream wording.

Unmeasured, and why: the plugin has no outcome eval cases or probes, so the two skills ship on the two-labeler meaning diff alone under the owner's Q7 decision for this rollout (recorded on #6476), which relaxes the AGENTS.md eval-evidence rule for wording-only cuts. Descriptions are byte-identical, so triggering is untouched. The issue's "approve the eval spend" step does not apply (no eval ran; subscription usage is capacity, not cost).

Verification

• Meaning diff: docs-hygiene:compare-labeler, 2 runs per changed skill, before copy from git archive origin/main: boris and skill-authoring PASS. Seven rationale sentences the review flagged are restored (rationale is cut only when a case covers it).
• An invariant script compared frontmatter, fenced blocks, blockquotes, headings, link targets, URLs, backtick spans, numbers and quoted strings before and after: all changed files unchanged on those.
• boris/scripts/update.test.sh 42/42, skill-authoring/scripts/update.test.sh 32/32; skill-quality FAIL/WARN counts identical before and after for all five skills; fable-5 is byte-identical to main.
• Merged origin/main: its one playbooks change since the base (fix(evals): run invocation probes without subagents and name the run line in run-validity #11901, rewrite-a-plugin.md) touches no rewritten file.

Related

• Part of Concise skills: one skill-creation standard, evals, and a fleet-wide rollout #6476 (concise skills rollout).

🤖 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 2026-11-01T16:09:14.389059Z	94fa633	Draft marked ready

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.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 94fa633a1b

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[github-actions]

claude-security-review reviewed 033ad180e31899327819b8970242cc959baa4009: read 1 of 1 changed file(s); 0 finding(s), 0 as inline comments.

Security review (incremental since beab815): no issues found

This review covers only the change since beab815: one line in the changelog fragment .changes/playbooks/docs-6572-concise-skills-5b2e9d41.md.

What changed: the changelog entry no longer lists fable-5 among the rewritten skills. The PR description gives the reason: the fable-5 rewording was reverted to main's text. The entry now names only boris and skill-authoring, which matches the description.

Security surface: none.

• The fragment is release-note prose. No trusted-path glob matches it.
• It is not a standing instruction, so it removes no guardrail and the instruction-surface lens does not apply.
• It changes no plugin manifest (no manifest lines apply), no trust control, no token scope, no kill switch and no merge hold.
• The PR context contains no embedded instructions aimed at this lane.

The other 11 files in this PR are unchanged since the last reviewed commit and were reviewed then.

critical-findings: 0

[github-actions]

claude-review reviewed 033ad180e31899327819b8970242cc959baa4009: read 1 of 1 changed file(s); 0 finding(s), 0 as inline comments.

Incremental review (beab815 → 033ad18)

Scope: .changes/playbooks/docs-6572-concise-skills-5b2e9d41.md. I read all of its change.

What changed: The changelog fragment no longer lists fable-5 among the rewritten skills. It now names only boris and skill-authoring.

Check against the rest of the PR:

• The full diff (.claude-lane/pr.diff) changes no fable-5 file. The changed paths are under plugins/playbooks/skills/boris/ and plugins/playbooks/skills/skill-authoring/, plus this fragment. The fragment now matches what the PR ships.
• The PR body's decision table says the same thing: fable-5 is unchanged and its earlier rewording was reverted to main's text.
• bump: patch is unchanged. A patch bump still fits a wording-only change.

Closing issue #6572: The PR body has a per-skill decision table and names each unmeasured skill with its reason. That covers the issue's "Done when" items as far as this lane can check. The "measured" criterion rests on the owner decision linked from the PR body; the lane cannot check that decision.

Note (not a finding): The PR title still says "tighten three skill bodies", but the body and this fragment now say two. The title becomes the squash-commit subject, so the author may want to update it.

No findings in this increment.

[kyle-sexton]

Reply to the claude-review summary on 94fa633a1b:

#	Finding	Classification	Evidence	Reaction
1a	Rationale sentences cut with no case covering their behavior (rewrite-a-plugin.md#L43-L45)	VALID (fix now)	Fixed in beab815: all seven restored ("Scope shrinks; integrity doesn't.", "A comparison you will not act on is decoration.", both planning.md sentences, "Nothing real changes while the plan is still being corrected.", "The math: ...", and automation.md's "It never changes anything behind your back."). Only filler and intensifier deletions remain.	👍
1b	Done-when "every shipped description and body was measured" not met; the "wording-only rule" is not linked	INTENTIONAL	The owner's Q7 decision (2026-10-31T23:23Z, "go w/ recommended" to an option that stated it relaxes the AGENTS.md eval-evidence rule): wording-only rewrites ship on the two-labeler meaning diff; full measurement only for skills whose rules change. Recorded verbatim on #6476 (#6476 (comment)), linked from the PR body. The eval-spend step is moot: no eval ran, and subscription usage is capacity, not cost (owner standing direction).	👀