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 excellent operational skill body: phased workflow, executable commands, explicit validation and retry limits, and safety boundaries around pushing that anticipate real failure modes (rejected pushes, unpushed user commits, flaky reappearing failures). The only structural question is whether the triage and pitfalls tables warrant a reference file as the skill grows.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and dense — triage and pitfalls are compact tables, commands are bare, and the prose only carries non-obvious operational judgment (detached-HEAD scoping, @{u}..HEAD semantics, rejected-push policy) that Claude does not reliably know. Every token earns its place. | 5 / 5 |
Actionability | Fully executable throughout: 'gh run view <run-id> --log-failed', 'gh run list --branch "$(git branch --show-current)" --status failure --limit 5', 'git log --oneline @{u}..<start commit>', 'gh run watch <databaseId> --exit-status' — copy-paste-ready commands with concrete fallbacks for each input form. | 5 / 5 |
Workflow Clarity | Clear three-phase sequence with explicit validation (verify locally with the same command from the workflow YAML before committing), feedback loops (revert and retry max 3 attempts; re-diagnose after pushed runs fail, max 2 more rounds), and a Rules checklist. The batch operation is fully covered by validation, so no cap applies. | 5 / 5 |
Progressive Disclosure | Well-organized single-file structure with clear sections and no unnecessary external references, and nothing inlined that clearly belongs in a separate file — but at ~90 lines it exceeds the under-50-line simple-skill exception, and the triage/pitfalls tables are borderline reference material. Sits between anchors 4 and 5. | 4 / 5 |
Total | 19 / 20 Passed |