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 exemplary instruction-style skill: lean, guardrail-dense, fully executable main path with copy-paste evidence capture, and a clearly sequenced choose→execute→report workflow with explicit validation checkpoints. The only weakness is that its two cross-skill references point outside the bundle and cannot be verified within it.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean, imperative, and assumes Claude's competence — e.g. "Never substitute a remembered count for this inventory", "An alias, workflow job, matrix instance and test case are different units" — with no explanation of concepts Claude already knows. Every sentence is a guardrail or instruction, matching anchor 5 ("every token earns its place") rather than 4, which requires over-explanation that could be trimmed. | 5 / 5 |
Actionability | Concrete commands are named ("npm run check:gate-manifest", "npm run quality:scan", "npm run check:quality-skill-contract") and the bash block with mktemp, exit-code capture, and log path is copy-paste ready and covers the common execution case, matching anchor 5. Vague spots like "the gate process supervisor when the candidate provides it" are conditional fallbacks, not gaps in the main path. | 5 / 5 |
Workflow Clarity | The three sections form a clear sequence (choose scope → execute and preserve evidence → report) with explicit validation checkpoints: capture "the command's own exit before displaying a summary", "Missing dependencies, timeout, cancellation and unavailable artifacts are incomplete evidence, not PASS", and error-recovery direction via "use the causal comparison in the shared base-green protocol". This matches anchor 5; the batch/destructive cap does not apply since the scan is read-only and exit-code interpretation is explicitly validated. | 5 / 5 |
Progressive Disclosure | The single-file body is well organized with clear section headers and clearly signaled one-level-deep references ("Read `../_shared/base-green.md` before interpreting base or PR health", "use `../validate-release-green/SKILL.md` for the local observer's limited scope"). However those referenced sibling files are not part of the bundle (no references/, scripts/, or assets/ directories exist), so the skill depends on unverified external context — a minor organization gap that keeps it below anchor 5's fully self-contained, easy navigation. | 4 / 5 |
Total | 19 / 20 Passed |