Content
67%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 rich, opinionated design skill with unusually concrete guidance (exact values, tools, timings, and self-tests) and a clean one-level-deep reference structure. Its main weakness is token efficiency: three sections recap the same insights and two trailing sections are non-instructional. The design workflow is sequenced and checkable but split across redundant sections rather than one integrated loop.
Suggestions
Merge the three overlapping recap sections — 'Design Process', 'Implementation Notes', and 'Quick Diagnostic' — into a single workflow, since 'Implementation Notes' repeats 'Design Process' nearly verbatim and the diagnostic table restates the same key insights a third time.
Cut 'About the Author' and 'Further Reading' (affiliate-tagged book links) from SKILL.md; neither instructs the agent, and the agency background already belongs in references/case-studies.md.
Trim or fold each pillar's 'Why it works' rationale paragraph into 'Key Insights', and make the feedback loop explicit in the workflow (score → apply diagnostic fixes → re-score) so validation is an ordered step rather than an implied one.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Core guidance is dense and specific (exact cubic-bezier curves, ms timelines, 10:1 type ratios, hex variants), but 'Implementation Notes' repeats the 'Design Process' section nearly verbatim, the 'Quick Diagnostic' recaps the same insights a third time, and 'About the Author' plus affiliate-tagged 'Further Reading' add no instruction. Mostly efficient but could lose roughly 30% of tokens without losing guidance. | 3 / 5 |
Actionability | Highly concrete directives throughout — 'cubic-bezier(0.16, 1, 0.3, 1) (expo out)', 'hero title words staggered (200-600ms, 80ms stagger)', 'minimum 10:1 (180px headline / 14px body)', '#0a0a0a', 'Use Lenis or Locomotive Scroll', 'Lighthouse, targeting 90+' — plus actionable self-tests (blur test, squint test, screenshot test). Falls short of 5 because a meaningful share of guidance is aspirational rather than executable ('magnetic compositions, elements that breathe and surprise', 'typography that makes you gasp') and complete copy-paste examples live in references rather than inline. | 4 / 5 |
Workflow Clarity | A clear sequenced 'Design Process' (Concept → Signature moment → Typography → Motion → Ship with restraint) with a built-in feedback loop in the scoring section ('Always provide the current score and specific improvements needed to reach 10/10') and a 12-question diagnostic checklist with fixes. Not a 5 because the process is duplicated across two overlapping sections and the score → fix → re-score iteration is implied rather than an explicit ordered step. | 4 / 5 |
Progressive Disclosure | Each pillar ends with a well-signaled pointer ('See: [references/typography.md] for font pairing strategies') plus a consolidated 'Reference Files' section; all five referenced files exist, are one level deep with no nested references, and contain real content. Not a 5 because the ~440-line body itself carries content (category scoring rubric, common mistakes, agency bios) that could partly live in the references, stretching the 'overview' role. | 4 / 5 |
Total | 15 / 20 Passed |