Content
92%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 skill body: a lean skeleton with exact gh/git commands, explicit safety and confirmation gates, and clean one-level delegation to a well-organized rules file. The only minor improvement would be supplying concrete commands for the duplicate-search step.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and entirely directive with no explanation of concepts Claude already knows — every line (Input handling, Hard Rules, numbered state-establishment, the confirm gate) earns its place. | 5 / 5 |
Actionability | Core operations are copy-paste ready ('gh issue view --comments', 'gh pr list --search "<number>"', 'git fetch origin main', 'gh api rate_limit'), but a few steps remain high-level hints such as 'search open and closed issues and PRs for the same symptom' without a specific command. | 4 / 5 |
Workflow Clarity | A clear numbered two-phase sequence with explicit validation checkpoints: the stale-checkout guard ('A stale checkout produces "already fixed" mistakes'), the rate-limit check before sweeps, and the confirmation gate before posting, labeling, or closing — destructive actions are validated, so the cap at 3 does not apply. | 5 / 5 |
Progressive Disclosure | The body is a concise overview that delegates all detail through one clearly-signaled, one-level-deep link (references/rules.md) whose scope is enumerated ('safety, the parity check, reproducing, the verdict and the reply'); the referenced file exists and does not chain into further nested skill docs. | 5 / 5 |
Total | 19 / 20 Passed |