Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a strong, highly actionable skill with excellent workflow clarity and explicit safety guardrails. The 12-step process is well-sequenced with concrete commands, validation checkpoints, and clear error recovery paths. The main weaknesses are some redundancy in safety warnings across multiple sections and the monolithic structure that could benefit from splitting auxiliary guidance into referenced files.
Suggestions
Consolidate the repeated safety rules (no push to main, no git add -A, no git clean -fdx) into a single Safety section rather than repeating them across Safety, Known Footguns, and Guardrails.
Consider extracting the changelog and docs update guidance (steps 5-6) into a referenced file to reduce the main skill's length and improve progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is mostly efficient and avoids explaining concepts Claude already knows, but there is some redundancy—safety rules about not pushing to main and not using `git add -A` are repeated in Safety, Known Footguns, and Guardrails sections. The overall length is justified by the complexity of the workflow, but could be tightened by consolidating repeated warnings. | 2 / 3 |
Actionability | The skill provides fully executable shell commands at every step, with specific flags, variable assignments, and concrete examples. Commands are copy-paste ready with clear placeholders like `<PR>`, and fallback paths are provided (e.g., if `committer` is not found, use `git commit`). | 3 / 3 |
Workflow Clarity | The 12-step workflow is clearly sequenced with explicit validation checkpoints: conflict resolution limits (stop after 3), gate retry limits (max 3 cycles), mandatory verification that PR is not behind main after push, and a final gate-pass check before printing the completion message. Feedback loops for error recovery are well-defined throughout. | 3 / 3 |
Progressive Disclosure | The content is well-structured with clear sections and headers, but it's a long monolithic file with no references to supporting documents. The review findings dependency on `.local/review.md` is well-signaled, but content like the changelog update process or docs update guidance could potentially be split out. For a skill of this complexity, the single-file approach is borderline acceptable but not ideal. | 2 / 3 |
Total | 10 / 12 Passed |