Content
86%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, tight skill body: fully executable commands throughout, a genuine local-validation gate, and clean organization appropriate to its size. The only trims are the redundant Description section and motivational filler, plus making the push-workflow ordering and gating explicit rather than implied.
Suggestions
Delete the '## Description' section — it restates the frontmatter description verbatim in function and adds no new information.
Replace commentary like 'Do not waste the tokens or the compute' with the operative instruction ('If local checks fail, fix before pushing — do not push'), and state explicitly that local checks passing is a gate for pushing.
Order the rules as the actual push workflow (check in-flight runs → run local checks → push → cancel superseded → monitor) so the sequence reads top-to-bottom.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean — rules are stated as directives with commands, no explanation of concepts Claude already knows — but a few tokens are not fully earned: the "## Description" section ("Guidelines for agents to minimize GitHub Actions compute waste...") duplicates the frontmatter, and phrases like "Do not waste the tokens or the compute" are motivational padding. This fits the 'efficient; minor instances of over-explanation that could be trimmed' anchor. | 4 / 5 |
Actionability | Every rule ships a copy-paste-ready command: `gh run list --branch $(git branch --show-current) --limit 5`, `gh run cancel <run_id>`, `./gradlew spotlessApply spotlessCheck detekt assembleDebug test allTests`, `gh run view <run_id>`, and `gh run rerun --failed`. Fully executable guidance covering the common cases, matching the top anchor. | 5 / 5 |
Workflow Clarity | A clear implied sequence exists (check in-flight runs → run local checks → push → cancel superseded) with a real validation checkpoint (the local gradle gate before pushing). It falls short of 5 because the sequence is scattered across numbered rules rather than ordered, and there is no explicit 'only push when local checks pass' gate or post-push monitoring loop; it sits above 3 because the local-test checkpoint and cancel-superseded rule are explicit. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines, single-purpose, and needs no external references; content is well-organized into numbered rules plus a Monitoring section with no buried or nested references. Per the rubric's simple-skill guidance, this earns the top score on well-organized sections alone. | 5 / 5 |
Total | 18 / 20 Passed |