Content
77%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 disciplined router: a sequenced five-phase workflow with real validation gates, executable lint/detect/Lighthouse commands, and hard numeric build rules. Its one material defect is bundle integrity — the majority of the Layer A style skills it routes to, plus the whole ui-ux-db subtree, are absent, so the routing table promises files that do not resolve; secondary costs are duplicated descriptions of the reference library and one command targeting a missing script.
Suggestions
Restore or delete the missing referenced files — 15 design references (most Layer A taste skills, output-/stitch-/imagegen-*) and the entire ui-ux-db/ directory (README.md and scripts/search.py) are named in SKILL.md but absent from the bundle; every path in the routing table and Layer A/Layer B lists should resolve.
Remove or gate the ui-ux-db CLI commands (python3 -B $SKILL_DIR/references/ui-ux-db/scripts/search.py ...) until that script ships, since the command as written fails.
Trim the overlap between the routing table, the Layer A table, and "The reference library" section, which each describe the same files, and cut the Activation section that restates the frontmatter description.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, rule-first, and assumes Claude's competence — no explanations of concepts Claude already knows, and every line states a rule, threshold, or route. Not a 5: the routing table, Layer A table, and reference-library section each re-describe the same reference files, and the Activation section restates the frontmatter description — all trimmable duplication. | 4 / 5 |
Actionability | Gives copy-paste-ready commands ("bun $SKILL_DIR/scripts/perfection/check-draft-guard.ts", "detectPage({ kind: \"url\", url }, { reflow: true })", "uv run $SKILL_DIR/scripts/perfection/lighthouse-audit.py") and hard numeric thresholds (44x44, 24x24, 4.5:1, 1.25x, 65-75 chars). Not a 5: the ui-ux-db search CLI command points at references/ui-ux-db/scripts/search.py, which does not exist in the bundle, so that command is not actually executable. | 4 / 5 |
Workflow Clarity | A clear sequenced flow (ground -> brief -> build -> subtract -> verify) with explicit validation checkpoints and feedback loops: "Lint before visual QA" with "Exit 1 means a tell is in the draft: remove it", "Fix every error and warning before /visual-qa", a one-row-per-check WCAG 2.2 AA evidence table, and a bounded review loop. Matches the top anchor with checklist and error-recovery loops. | 5 / 5 |
Progressive Disclosure | The router design itself is exemplary — a request-to-reference table, one-level-deep references, and "read exactly those" discipline — but scored against the actual bundle, 15+ named files are missing: most Layer A taste skills (taste-skill.md, minimalist-skill.md, soft-skill.md, gpt-tasteskill.md, brutalist-skill.md, redesign-skill.md, image-to-code-skill.md, output-skill.md, stitch-skill.md), all imagegen-* files, and the entire ui-ux-db/ directory. Not a 4: navigation through the core Layer A routing breaks for most named paths, which is more than a minor organization gap. | 3 / 5 |
Total | 16 / 20 Passed |