Conciseness

The body is efficient and every non-obvious fact (the .source submodule, diff.ignoreSubmodules=all, lint-staged/biome hooks, gitlink bump rules) earns its tokens with no general-concept padding. It is not 5 because the workflow sequence is restated three times (intro paragraph, checklist, step sections) and could be tightened; not 3 because there is no explanation of concepts Claude already knows.

Actionability

Guidance is fully executable: exact commands such as 'git worktree add -b review/<branch>-<BASE_SHA:0:7> ../review-<BASE_SHA:0:7> <BASE_SHA>', 'git -c diff.ignoreSubmodules=none commit', a complete copy-paste subagent prompt template, a HEREDOC commit message, and the ff-only merge plus teardown commands. It is not 4 because commands and templates cover the common cases end-to-end with only intentional SHA/branch placeholders.

Workflow Clarity

The five steps are clearly sequenced with explicit validation checkpoints and error-recovery loops: capturing BASE_SHA as the sole review target, 'Verify before deleting' with Grep, ReadLints on touched files, 'git log --oneline -3' history confirmation, the --ff-only divergence handling ('do not force anything; report the divergence'), the worktree-path-exists abort fallback, and the clean-result path when the review finds nothing. It is not 4 because both the checkpoints and the failure paths are explicit rather than implicit.

Progressive Disclosure

No bundle files exist and the single ~155-line file is organized into clear sections (Submodule, Workflow, Guardrails, Related skills) with well-signaled links to sibling skills. It is not 5 because the inlined ~25-line subagent prompt template is reference-shaped content that could live in a separate file and the body exceeds the simple under-50-line case; not 3 because nothing is buried and the structure is genuinely good.

Specificity

The description enumerates the full pipeline concretely — 'Commit local changes as a baseline', 'hop into a lightweight review worktree', 'run a thermo-nuclear code quality review scoped to ONLY that commit', 'commit the resulting refactor as a separate follow-up commit', 'land it back and tear the worktree down', plus submodule parking — matching the comprehensive-coverage anchor. It is not 4 because the action list covers the entire workflow with no meaningful gaps.

Completeness

It clearly answers 'what' (the commit → worktree → scoped review → refactor commit → land/teardown sequence) and 'when' via the explicit 'Use when the user asks to review their local/uncommitted changes for AI slop and redundant code, or invokes nv-park-and-review' clause with concrete trigger phrases. It is not 4 because the 'when' is already explicit and specific rather than improvable.

Trigger Term Quality

Natural phrases like 'review their local/uncommitted changes', 'AI slop', 'redundant code', and the explicit 'invokes nv-park-and-review' give good keyword coverage, but a few natural user phrasings ('clean up my diff', 'code quality review of my changes') are absent. It is not 5 because synonym coverage is not comprehensive, and not 3 because the key natural terms users would actually say are present.

Distinctiveness Conflict Risk

The park/worktree/land-commit pipeline is a distinct niche, but the trigger 'review their local/uncommitted changes for AI slop and redundant code' overlaps with closely related skills such as deslop and thermo-nuclear-code-quality-review. It is not 5 because of that minor overlap risk, and not 3 because the workflow framing (baseline commit, worktree, follow-up refactor commit) is clearly distinguishable from a plain review skill.