Content
96%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
An exemplary instruction-only skill body: lean, entirely repo-specific, with concrete seams, an ordered review workflow, explicit stop conditions, a merge bar, and example comments. The only notable gap is that the inline checklist duplicates content delegated to the authoritative REVIEW.md, which slightly muddies what lives where.
Suggestions
Clarify the division of labor with REVIEW.md: either state that the inline 'Checklist focus' is a summary and REVIEW.md holds the authoritative version, or trim the inline checklist to only the highest-frequency checks to reduce duplication.
Add one or two exact verification commands (e.g., how to run the layer-appropriate test suites or guard tests) so reviewers can validate PR claims like 'tests pass' directly instead of only judging the listed evidence.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with repo-specific rules Claude cannot know (dependency boundaries, StageNames guards, manifest paths) and contains no padding or explanation of already-known concepts — every token earns its place, matching the lean top anchor rather than the 'minor instances of over-explanation' of 4. | 5 / 5 |
Actionability | Guidance is fully concrete: exact file paths and seams ('Trackdub.Domain', 'IRuntimePlanner', 'StageNames.*', 'bundled-models.manifest.json', 'tests/Trackdub.TestDoubles/') plus five copy-paste-ready example review comments covering common cases. Per the rubric's instruction-only note, absence of code is not penalized when guidance is this actionable, so it exceeds 4's 'minor gaps'. | 5 / 5 |
Workflow Clarity | The workflow is clearly sequenced — required reading (1-3), issue-context loading with explicit fallback branches, then review priorities in order 1-7 — with explicit validation gates ('Automatic review stops', 'Merge bar') and checklists ('Checklist focus'), matching the top anchor's checkpoints and error recovery; 4's 'minor validation gaps' does not apply. | 5 / 5 |
Progressive Disclosure | Sections are well organized and references are clearly signaled and one level deep ('Root REVIEW.md (authoritative reviewer checklist)', 'AGENTS.md', '.github/pull_request_template.md'), with no nested references. However, the inlined 'Checklist focus' partially duplicates the external file the skill itself calls authoritative (REVIEW.md), leaving a minor organization gap versus the 5 anchor's 'content appropriately split'. | 4 / 5 |
Total | 19 / 20 Passed |