Content
88%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 unusually actionable skill body: exact motion values, frequency/purpose gating with explicit verdicts, a sequenced workflow with completion criteria, and a required output format that forces showing rejections. Its only weaknesses are minor editorial padding and no externalization of reference-style detail.
Suggestions
Trim meta-commentary sentences ('This section is what separates this skill from an animation wishlist', 'A search skill. It does ONE thing...') — the rules already imply them, saving tokens without losing guidance.
If the skill grows, move the duration-budget table, easing-token vocabulary, and 'Where to Hunt' catalog into a references/ file linked one level deep, keeping SKILL.md as the gate + workflow overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational — budget tables, exact CSS/spring values, grep sweep patterns — with only minor trimmable editorial lines such as 'This section is what separates this skill from an animation wishlist' and 'A search skill. It does ONE thing...'. Not 5: a handful of meta-commentary sentences don't earn their tokens; not 3: padding is minor and localized. | 4 / 5 |
Actionability | Guidance is fully concrete and copy-paste ready: 'transform: scale(0.97)' with 'transition: transform 160ms ease-out', 'clip-path: inset(0 100% 0 0)' hold-to-confirm specs, spring configs '{ type: "spring", duration: 0.5, bounce: 0.2 }', velocity threshold '~0.11', named easing curves, grep patterns, and a filled-in example output table. Not 4: common cases are covered with exact values, not just templates. | 5 / 5 |
Workflow Clarity | Recon → Sweep → Gate → Report is clearly sequenced with explicit checkpoints: 'Done when every seam class has either yielded candidates with file:line evidence or been explicitly cleared', a four-question gate that 'must survive all four questions, in order' with recorded answers, and a required rejected-candidates section. The skill is read-only, so the destructive/batch validation cap does not apply. Not 4: checkpoints (done-when, ordered gate, mandatory rejections) are explicit, not implicit. | 5 / 5 |
Progressive Disclosure | A single well-organized file with clear section headers and easy navigation; no bundle files exist. Not 5: at ~135 lines, detail such as the duration/easing tables and the 'Where to Hunt' catalog could live in a one-level-deep reference; not 3: everything inline is genuinely used in a single pass and nothing is buried. | 4 / 5 |
Total | 18 / 20 Passed |