Content
81%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 exceptionally actionable, well-sequenced procedural skill: copy-paste commands, explicit gates, and genuine feedback loops at every risky step. Its weaknesses are verbosity — repeated policies and over-long monolithic paragraphs — and inline detail that could be split out for easier navigation.
Suggestions
Deduplicate repeated policy statements (e.g. 'a push does not launder its own contents', credential stripping) by stating each once and cross-referencing the section.
Split the Phase 4 test-check bullet into labeled sub-steps (environment prep, publish in-progress, execute, publish result, drift handling) to match the structure of the other phases.
Move stable templates (the test-report comment structure, the published-body skeleton) into a reference file loaded on demand, keeping SKILL.md's core workflow leaner.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with operational instruction and contains no concept-explainer filler, but it can be tightened: 'a push does not launder its own contents' appears twice, credential-stripping instructions are restated across sections, and Phase 4's test-check bullet is a single multi-hundred-word paragraph mixing planning, publishing, drift checks, and fallbacks. Anchor 3 ('mostly efficient but could be tightened') fits better than 4, whose 'minor instances' understates the repetition. | 3 / 5 |
Actionability | Fully executable guidance throughout: exact gh commands with --json/--jq filters ('gh api .../reviews --paginate --jq ".[] | select(.user.login == \"<factory-app>[bot]\") | {state, submitted_at, body}"'), credential-stripped test commands ('env -u GH_TOKEN -u GITHUB_TOKEN pnpm --filter <pkg> test'), exact body-file paths, verdict-line gates, and fallback procedures. Placeholders are only where runtime values must be substituted. Matches anchor 5. | 5 / 5 |
Workflow Clarity | Phases 1–7 are clearly sequenced with explicit validation checkpoints and feedback loops: the body-file gate before posting, the head-not-moved check with re-verification on drift, the in-run cross-check with a single re-run-and-retry on mismatch, the rejected-verdict retry loop, and the mandatory adversarial check before every approve. This is exactly anchor 5's 'clear sequence with explicit validation steps and feedback loops'. | 5 / 5 |
Progressive Disclosure | The skill has no bundle of its own and delegates category guidance one level deep to the factory-review skill's references/categories/README.md and references/archaeology.md with exact load instructions — clearly signaled. Sections are well-organized with headers, but long procedural detail (the test-report format spec, author-identity misconfiguration handling) is inlined monolithically, leaving minor organization gaps — anchor 4 rather than 5. | 4 / 5 |
Total | 17 / 20 Passed |