Content
73%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 an unusually crisp, opinionated review playbook with concrete thresholds, a clear remedial hierarchy, and explicit verdict criteria — its workflow and actionability are strong. Its one serious defect is that the entire detailed standards catalog it repeatedly cites lives in a STANDARDS.md file that is absent from the bundle, leaving the progressive-disclosure promise broken and exact values unretrievable.
Suggestions
Ship STANDARDS.md alongside SKILL.md (or inline the essential easing/duration/spring tables) — the body defers exact values to it four times, but no such file exists in the bundle.
Inline a few fallback defaults (e.g., a recommended cubic-bezier and per-element duration budgets) so findings can cite specifics even when STANDARDS.md is unavailable.
Trim the redundant second 'Examples' paragraph and the review-method provenance paragraph; both restate scope or background already covered elsewhere.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes Claude's competence — it never explains what CSS transitions, keyframes, or Framer Motion are, and nearly every line is a rule or threshold ('UI animations stay under 300ms', 'Keyboard-initiated and 100+/day actions get **no** animation'). Minor trimmable padding remains: the second 'Examples' paragraph restates scope already covered in 'When to Use' and 'Limitations', and the attribution paragraph about the review method's provenance is background flavor, keeping it below the lean anchor at 5. | 4 / 5 |
Actionability | Guidance is highly concrete — specific thresholds ('under 300ms', 'scale(0.9–0.97) + opacity', '30–80ms stagger') and copy-paste fixes in the findings table ('transition: transform 200ms ease-out', 'var(--radix-popover-content-transform-origin)'). It misses 5 because exact values (curves, duration tables, spring configs) are deferred to 'pull the exact one from [STANDARDS.md] rather than approximating', and STANDARDS.md does not exist in the bundle, so the promised precision cannot be retrieved. | 4 / 5 |
Workflow Clarity | This single-purpose skill has an unambiguous sequence: measure every animation against the ten standards, flag escalation triggers on sight, propose fixes via the nine-step remedial hierarchy, then emit Part 1 (findings table) and Part 2 (tiered verdict) with explicit Block/Approve criteria that function as validation checkpoints for the decision. The destructive/batch validation cap does not apply since this is a read-only review skill, and the explicit decision criteria ('Block — any feel-breaking regression…', 'Approve — no feel-breaking regressions…') leave no ambiguity about the single action. | 5 / 5 |
Progressive Disclosure | The in-body structure is good and the reference is clearly signaled and one level deep ('For the full rule catalog (easing curves, duration tables, spring config, gestures, clip-path, performance, a11y), see STANDARDS.md'), but scoring against the actual bundle, STANDARDS.md does not exist — there are no references/, scripts/, or assets/ directories at all. Four references to a missing file make the skill's deferred detail catalog a dead end, which is a fundamental organization failure rather than a mere clarity gap. | 2 / 5 |
Total | 15 / 20 Passed |