Content
92%Weight 40%Scale 1-3Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is lean, actionable, and well-sequenced with strong validation checkpoints. Its main weakness is reference depth: the command index points through a group index to per-subcommand files, adding an extra hop of indirection.
Suggestions
Flatten the reference hierarchy so the body links directly to group index files (e.g. references/pr.md) and have each group index contain its subcommand details inline, reducing commands.md → group → subcommand to a single level of indirection.
In the Reference section, state explicitly that commands.md is only an index and that group files (pr.md, issue.md, etc.) are the primary entry points, so Claude does not always load two files before reaching guidance.
Consider linking the most-used subcommand files (e.g. pr-review.md, pr-diff.md) directly from the relevant Operation Decision bullet to skip the intermediate index hop for common tasks.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean body with no concept-explanation padding; every section is an actionable rule or checklist that assumes Claude's competence, with each token earning its place. | 3 / 3 |
Actionability | Provides concrete commands and flags (`gh api --paginate`, `gh api -f/-F`, `gh api repos/{owner}/{repo}/pulls/{num}/comments`) plus specific gotchas (422 from mismatched `commit_id`/`headRefOid`), giving executable guidance rather than vague direction. | 3 / 3 |
Workflow Clarity | Sequenced Workflow steps, an Operation Decision branch table, Failure Handling, and an explicit pre-write Safety Checklist with validation checkpoints and idempotency handling for batch/destructive operations. | 3 / 3 |
Progressive Disclosure | Overview body signals references clearly, but navigation is two levels deep (commands.md → group index like pr.md → subcommand pr-diff.md), exceeding the one-level-deep ideal in the rubric anchor. | 2 / 3 |
Total | 11 / 12 Passed |