Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is highly actionable with concrete gh commands and a clear sequenced workflow, scoring well on actionability and workflow clarity. It is held back by mild redundant editorializing and a monolithic structure with no progressive disclosure to reference files.
Suggestions
Tighten the intro and remove restated philosophy ('This skill performs a deep read...', 'do not just summarize the text') that duplicates the description to lift conciseness toward 3.
Move the Output Template and/or the detailed flag/analysis logic into a references/ file (e.g., TEMPLATE.md) and link to it from the body to apply progressive disclosure and reduce inline length.
Add a brief error-recovery note (e.g., what to do if 'gh auth status' fails or a rate limit is hit) to make the workflow's checkpoints fully explicit.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is largely command-dense and lean, but the opening paragraph re-states the 'deep read' philosophy already conveyed in the description, and coaching such as 'do not just summarize the text. Perform a logic check:' could be tightened. Not the level-3 ideal where every token earns its place. | 2 / 3 |
Actionability | It provides concrete, executable commands with exact JSON field lists (e.g., 'gh issue list --repo <owner/repo> --state open --json number,title,body,author,...'), 'gh pr diff <number>', and 'gh pr view <number> --json mergeable,statusCheckRollup', plus a copy-paste output template. Placeholders like <owner/repo> are the only needed substitutions. | 3 / 3 |
Workflow Clarity | A clear numbered four-step sequence (Initialize Mental Model, Retrieve State & Delta, Deep Analysis, Pattern Synthesis) with explicit checkpoints — an auth prerequisite, a mergeability verification, and a security keyword scan. The skill is read-only, so the destructive/batch feedback-loop cap does not apply. | 3 / 3 |
Progressive Disclosure | Content is well-sectioned rather than a wall of text, but it is monolithic: no references/, scripts/, or assets/ bundle files exist and the body references none, while the ~30-line Output Template and detailed flag logic sit inline. Over 50 lines, so the simple-skill carve-out does not apply. | 2 / 3 |
Total | 10 / 12 Passed |