Content
82%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 well-crafted, highly actionable instruction-only skill with concrete flag patterns, a remedial hierarchy, and an enforced output format. Its main weaknesses are redundancy between the standards and escalation-triggers lists and a referenced STANDARDS.md that is absent from the bundle, leaving all precise values (curves, durations, spring configs) unreachable.
Suggestions
Ship STANDARDS.md alongside SKILL.md — the body delegates all exact values and per-element budgets to it, so a dangling reference cripples the skill at runtime and caps progressive disclosure.
Dedup the Ten Standards and the Aggressive Escalation Triggers (ease-in, scale(0), >300ms, layout properties, symmetric timing, hover/reduced-motion each appear in both) to tighten the token budget.
State the operational sequence explicitly (e.g. '1. Scan the diff for the escalation triggers, 2. Measure remaining motion against the ten standards, 3. Propose fixes from the hierarchy, 4. Emit the two-part output') so the workflow doesn't rely on section order alone.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes competence — it never explains what easing or transform-origin are, and every section carries rules or values. Minor redundancy exists: the Ten Standards and the Escalation Triggers repeat the same items (ease-in, scale(0), >300ms, layout properties, symmetric timing, hover gating, reduced-motion), and the Operating Posture persona rhetoric could be trimmed. That puts it at 'Efficient; minor instances of over-explanation that could be trimmed' rather than the every-token-earns-its-place 5. | 4 / 5 |
Actionability | For an instruction-only skill, the guidance is fully concrete: exact patterns to flag ('transition: all', 'transform: scale(0)', 'transform-origin: center'), a copy-paste-ready findings table with before/after fixes and rationale, a frequency-to-motion mapping, a remedial hierarchy ordered by preference, and explicit block/approve criteria with file:line citation requirements. Specific examples cover the common cases, matching the top anchor. | 5 / 5 |
Workflow Clarity | The review process is well sequenced implicitly — measure against ten standards, escalate on triggers, propose fixes via the remedial hierarchy, emit the two-part required output, close with an explicit verdict — and the output format functions as a validation checkpoint. It falls short of a 5 because the operational sequence is implied by section order rather than stated as an explicit step order (e.g. how to walk the diff, when to load STANDARDS.md relative to finding issues). This is a review skill, not a destructive/batch operation, so the validation-cap does not apply. | 4 / 5 |
Progressive Disclosure | Structure is strong: a concise overview with the full rule catalog correctly split out to a clearly signaled, one-level-deep reference ([STANDARDS.md], referenced three times with explicit load conditions — 'whenever a finding needs a precise value'). However, STANDARDS.md does not exist in the provided bundle (the directory contains only SKILL.md), so the only reference is dangling — more than a purely minor gap, which keeps it below 5 despite excellent in-file organization. | 4 / 5 |
Total | 17 / 20 Passed |