Content
85%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: verified copy-paste commands, a priority-ordered decision table, explicit zero-result handling with fallback labeling, and destructive-write safeguards, with clean one-level-deep progressive disclosure into two real reference files. The one real weakness is redundancy — the query-discipline rules are stated three times across Query Contract, the 0-results section, and Tips.
Suggestions
Delete the opening 'Searchable local UI/UX guidance: 79 searchable styles (50 active), 192 product palettes...' line — it duplicates the frontmatter description verbatim and adds nothing the priority table and domain tables don't already convey.
Merge the 'If a search returns 0 results' section into the Query Contract's existing retry rules (they repeat each other), keeping only the unique parts: the explicit built-in-defaults fallback and the 'never present a 0-result search as if it returned data' rule.
Cut the first three bullets of 'Tips for Better Results' ('one dominant intent and 2–5 meaningful terms', 'Retry once', '--design-system vs --domain') — all three restate the Query Contract — and keep only the troubleshooting table, which adds genuinely new pointers.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with commands and tables and explains no basic concepts, but ~20 lines are redundant: the opening inventory line repeats the frontmatter description verbatim, the "If a search returns 0 results" section restates Query Contract's retry/fallback rules, and the first three Tips bullets duplicate "one dominant intent, 2–5 meaningful terms" and "retry once" guidance already given. This fits 'mostly efficient but includes some unnecessary explanation or could be tightened' better than 4's 'minor instances', since three separate sections repeat the same query discipline. | 3 / 5 |
Actionability | Fully executable throughout: exact copy-paste commands with the absolute "${CLAUDE_PLUGIN_ROOT}/.claude/skills/ui-ux-pro-max/scripts/search.py" path, per-domain example queries, a full stack list, the dial table, and a worked example workflow. Every documented flag (--design-system, --persist, --page, --force, --variance/--motion/--density, -f, --json) was verified against scripts/search.py and matches. | 5 / 5 |
Workflow Clarity | Steps 1→4 are clearly sequenced with explicit validation and feedback loops: retry-once with narrower rewrite on empty/off-topic results, then fall back to the priority table while explicitly labeling output as built-in defaults, and "never present a 0-result search as if it returned data". Destructive operations are guarded (existing MASTER.md is skipped unless --force; read it before --force; never force without user authorization), and checklists exist (priority table, Pre-Delivery Checklist reference). Not 4 because checkpoints, error-recovery loops, and checklists are all explicit rather than merely 'mostly present'. | 5 / 5 |
Progressive Disclosure | Scored against the actual bundle: the body is an overview whose bulk data (119 UX guidelines, palettes, icons, presets) is reachable via the search tool rather than inlined, and only two references exist (references/quick-reference.md, references/pro-rules.md) — both real files, one level deep, clearly signaled with on-demand read instructions and specific, verified section pointers (§3–§9, named tokens like color-dark-mode and spring-physics all confirmed present). | 5 / 5 |
Total | 18 / 20 Passed |