Content
81%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.
A strong diagnostic body with exceptional actionability — symptom tables, concrete thresholds, executable snippets, and a validated three-phase workflow. Its main weakness is token efficiency: persona framing, an anecdote section, a scripted greeting, and re-explanations of rendering basics Claude already knows dilute genuinely non-obvious material such as the Motion version behavior and inherited CSS variable trap.
Suggestions
Remove the 'Initial Response' scripted greeting, the 'Why This Matters' anecdote, and the 'Operating Posture' persona narrative, keeping only the operative bar (mid-range phone as the target device, 8ms budget on 120Hz).
Cut explanations of rendering fundamentals Claude already knows (layout/paint/composite definitions, why transform avoids re-measurement, CSS variable inheritance, rAF running on the main thread) and keep only the non-obvious specifics: motion 14.2 shorthand acceleration, the inherited CSS variable trap, the sub-20px blur limit in Safari, and the box-shadow to drop-shadow swap.
Move the case studies (Vercel's tab indicator, Vaul's drawer drag, Sonner) and the Motion version-compatibility details into a references/ file linked from the relevant sections, shortening SKILL.md and adding well-signaled one-level-deep references.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient — the symptom table, tier/swap tables, and version-specific Motion details are dense and non-obvious — but there is more than minor over-explanation: a scripted 'Initial Response' greeting, the 'Why This Matters' anecdote, the 'Operating Posture' persona framing, and re-teachings of concepts Claude already knows ('Layout: calculate the size and position of every element', 'CSS variables inherit, so changing one on a parent triggers style recalculation for every child', 'JavaScript always runs on the main thread'). Fits the 3 anchor ('mostly efficient but includes some unnecessary explanation or could be tightened') better than 4, where over-explanation would be only minor. | 3 / 5 |
Actionability | Fully executable guidance throughout: a nine-row symptom-to-fix table, a concrete property-swap table (width/height to scale(), box-shadow to filter: drop-shadow()), copy-paste-ready jsx/js/css snippets, hard thresholds (16.7ms/8ms budget, blur under 20px, motion 14.2+), named tooling (DevTools Performance and Animations panels, remote debugging), and a structured report format with file:line examples. Specific examples cover the common cases, matching the 5 anchor. | 5 / 5 |
Workflow Clarity | A clear three-phase workflow (inventory animations in a table, diagnose via a likelihood-ordered checklist, report via a template) with explicit validation: 'Confirm with a recording, not by feel', Hard Rule 3 requiring findings to state code-verified vs device-recording-needed, and a 'Verified:' field in the report format. The checklist's final item ('Still shifting or stuttering? Now add will-change') acts as a fallback iteration. This is a read-only diagnostic skill, so the destructive/batch validation cap does not apply; the sequence matches the 5 anchor. | 5 / 5 |
Progressive Disclosure | No bundle files exist and the single body is well-sectioned, with the symptom table acting as an internal index ('Start here. Match what the user is seeing, then read the matching section'). However, at roughly 217 lines it exceeds the under-50-line simple-skill exception for scoring 5 on sections alone, and content such as the Vercel/Vaul/Sonner case studies and Motion version specifics could live in a one-level-deep reference file. Good structure with minor organization gaps fits the 4 anchor; not 3 since nothing is buried or clearly misplaced. | 4 / 5 |
Total | 17 / 20 Passed |