Content
100%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-only skill body: it is maximally lean, every rule is backed by a concrete example, edge cases (status reporting, break-the-rules conditions, pre-send checks) are explicitly handled, and the structure is clean for a single-file skill. No improvements needed.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every line instructs and the skill models its own compression — no concept explanations, no padding; even rule examples are minimal Bad/Good pairs. Matches anchor 5: lean, assumes Claude's competence, every token earns its place. | 5 / 5 |
Actionability | Each rule carries a concrete before/after example (e.g. 'Inline object prop, new ref, re-render. `useMemo`.'), `/caveman status` behavior is specified down to the no-hook fallback, and 'When to break the rules' enumerates specific cases. Per the scoring notes, an instruction-only skill with this concrete guidance is fully actionable. | 5 / 5 |
Workflow Clarity | A simple single-purpose skill whose single action (respond compressed, persist until 'stop caveman') is unambiguous; the 'Pre-send check' section acts as an explicit validation checkpoint before every send. The simple-skill exception applies and no destructive/batch cap is triggered. | 5 / 5 |
Progressive Disclosure | Roughly 45 content lines, no external references needed (the bundle contains no reference/script files), and clearly organized sections (Persistence, Floor, Rules 1-6, When to break, Pre-send check). Meets the under-50-lines simple-skill criterion for a top score. | 5 / 5 |
Total | 20 / 20 Passed |