Content
90%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.
The body is lean, executable, and well-sequenced with an explicit output contract, suiting a simple single-task skill. Its main gap is the lack of a failure-recovery loop and an unreferenced REFERENCE.md bundle file.
Suggestions
Add a brief failure branch, e.g. 'If it exits non-zero, read the stderr/diagnostics, fix the underlying issue, and re-run before finishing.'
Either reference references/REFERENCE.md from the body (e.g. 'See references/REFERENCE.md for evidence standards') or remove it from the bundle to keep navigation consistent.
Confirm the script path resolves at the skill's install location; the body points at .cursor/skills/.../scripts/check.sh while the bundle ships scripts/check.sh relative to SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely lean: a single sequenced instruction, one concrete command with its output contract, and a termination note. Every token earns its place and it assumes Claude's competence. | 5 / 5 |
Actionability | Provides a fully executable command with the exact expected stdout ('{"ok":true}') and failure semantics (exits non-zero), which is copy-paste ready for the single task. | 5 / 5 |
Workflow Clarity | A clearly sequenced single step with an explicit validation signal (exit non-zero on failure) and an ordering directive, but no feedback loop describing what to do when the check fails. | 4 / 5 |
Progressive Disclosure | Well-organized sections with a clearly signaled one-level-deep reference to scripts/check.sh; however references/REFERENCE.md exists in the bundle but is never referenced or linked from the body. | 4 / 5 |
Total | 18 / 20 Passed |