Content
65%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a well-organized, genuinely actionable set of 12 automatable UX review rules with concrete thresholds and named tooling. Its weaker spots are conciseness (some preamble/redundancy) and the lack of progressive disclosure or an explicitly sequenced end-to-end review workflow.
Suggestions
Tighten the Background & Scope section and collapse repetitive per-rule framing to reduce token overhead.
Add a short explicit review workflow (select relevant rules → run priority checks 9/12/1 → score pass/fail + severity → assemble report) with the report structure spelled out.
Consider moving the per-rule automation-detail into a single reference file (e.g., RULES.md) so SKILL.md serves as a lean overview pointing one level deep.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body avoids explaining basic concepts Claude already knows and is well-structured, but the Background & Scope preamble and the full per-rule Check/Automation/Fail-message layout across 12 rules carry framing and redundancy that could be tightened. | 2 / 3 |
Actionability | Each rule pairs a concrete "Check" with quantitative thresholds (e.g., "≥ 80%", "≤ 3 clicks", "tables > 20 rows", "WCAG 2.2 AA", "Lighthouse ≥ 90–95") and named tooling/scan targets (axe-core, WAVE, `aria-keyshortcuts`, `keydown`, `localStorage`, NLP similarity), giving specific actionable guidance appropriate for an instruction-only skill. | 3 / 3 |
Workflow Clarity | The Implementation Guidance gives a priority order, scoring scheme, tool-to-rule mapping, and skip conditions, but the end-to-end review process (how findings combine into the final structured report) is only lightly specified with no explicit checkpoints. | 2 / 3 |
Progressive Disclosure | The content is well-organized into clear sections and grouped categories, but it is a single self-contained file (~110 lines) with all rule detail inline and no one-level-deep references, so it functions more as a monolithic ruleset than an overview pointing to deeper materials. | 2 / 3 |
Total | 9 / 12 Passed |