Experimental · Built out of curiosity · Educational purposes · DYOR
Measure design quality. Don't guess.
The Lighthouse for design quality. 58 design intelligence commands with scoring algorithms, mathematical type scales, color palette generation, and anti-pattern detection.
Paste code → get a 0-100 score → install commands that fix what the score finds.
Most AI design skills tell the AI what good design looks like. OPTIK gives it scoring algorithms that measure design quality across five pillars:
| Pillar | Weight | What it measures |
|---|---|---|
| Typography | 25% | Scale ratio, banned fonts, hierarchy depth, line measure, weight usage |
| Color | 25% | Contrast ratios, palette coherence, anti-patterns, CSS variables, dark mode |
| Layout | 25% | Grid alignment, spacing rhythm, responsive breakpoints, nesting depth |
| Motion | 10% | Transition purpose, easing curves, reduced-motion support, performance |
| Accessibility | 15% | Focus styles, semantic HTML, ARIA, touch targets, screen reader experience |
The AI doesn't learn opinions. It runs analysis.
# All tools (auto-detects your AI harness)
npx skills add dragoon0x/optik
# Claude Code plugin
/plugin marketplace add dragoon0x/optik
# Manual
git clone https://github.com/dragoon0x/optik.git
cp -r optik/.claude your-project/ # Claude Code
cp -r optik/.cursor your-project/ # Cursor
cp -r optik/.agents your-project/ # Copilot/Kiro/Antigravity/OpenCode
cp -r optik/.gemini your-project/ # Gemini CLI
cp -r optik/.codex your-project/ # Codex CLIWorks with Claude Code, Cursor, Gemini CLI, Codex CLI, VS Code Copilot, Kiro, OpenCode, and Antigravity.
| Command | What it does |
|---|---|
/score |
Run full 0-100 design quality analysis |
/audit |
Comprehensive quality report with line references |
/critique |
UX hierarchy + resonance review |
/benchmark |
Compare against quality baselines |
/compare |
Side-by-side version comparison |
/report |
Generate shareable score card |
/trend |
Track score across iterations |
/certify |
Pass/fail against threshold (for CI/CD) |
| Command | What it does |
|---|---|
/type-scale |
Generate mathematical type scale from ratio |
/type-pair |
Font pairing recommendations |
/type-audit |
Audit all font usage across codebase |
/type-fix |
Auto-fix type scale inconsistencies |
/type-hierarchy |
Enforce heading + body hierarchy |
/type-measure |
Check line lengths (45-75 chars) |
/type-rhythm |
Align to vertical rhythm baseline |
/type-set |
Apply complete type system at once |
| Command | What it does |
|---|---|
/palette |
Generate full 50-950 shade palette from one hex |
/contrast |
WCAG AA/AAA contrast check |
/harmonize |
Fix palette coherence issues |
/semantic-color |
Generate semantic token set |
/dark-mode |
Create dark mode preserving contrast |
/color-audit |
Map all colors, find duplicates |
/color-fix |
Auto-fix contrast failures |
/color-a11y |
Ensure palette passes WCAG at every size |
| Command | What it does |
|---|---|
/grid |
Apply 8px grid with snap correction |
/spacing |
Generate spacing scale from base unit |
/density |
Audit visual density + whitespace |
/rhythm |
Enforce vertical spacing rhythm |
/responsive |
Check breakpoints + reflow quality |
/layout-audit |
Map layout patterns + find issues |
/layout-fix |
Auto-fix off-grid spacing values |
/align |
Verify alignment across sections |
| Command | What it does |
|---|---|
/animate |
Add purposeful motion + transitions |
/transition |
Audit + improve timing functions |
/scroll-fx |
Scroll-triggered reveals + parallax |
/motion-audit |
Check motion purpose + performance |
/reduce-motion |
Add prefers-reduced-motion support |
/overdrive |
WebGL, spring physics, GPU effects |
| Command | What it does |
|---|---|
/a11y-audit |
Full accessibility check |
/focus |
Generate focus ring system |
/touch-target |
Verify 44x44px minimum targets |
/aria |
Audit + fix ARIA attributes |
/a11y-fix |
Auto-fix detectable a11y issues |
/screen-reader |
Test screen reader experience |
| Command | What it does |
|---|---|
/tokens |
Generate complete token set |
/extract |
Pull reusable patterns into system |
/normalize |
Align to design system standards |
/system-audit |
Check for token drift + orphans |
/component-check |
Verify component consistency |
/export-css |
Export as CSS custom properties |
/export-tailwind |
Export as Tailwind config |
/export-json |
Export as W3C design tokens |
| Command | What it does |
|---|---|
/polish |
Final pass — alignment, spacing, details |
/distill |
Strip to essence, remove complexity |
/bolder |
Amplify safe, boring designs |
/quieter |
Tone down aggressive designs |
/harden |
Error handling, i18n, edge cases |
/delight |
Add memorable moments of joy |
| Score | Grade | Meaning |
|---|---|---|
| 90-100 | Exceptional | Ship with confidence |
| 80-89 | Strong | Minor refinements only |
| 65-79 | Needs work | Address issues before shipping |
| 40-64 | AI slop detected | Significant rework required |
| 0-39 | Design emergency | Start over with intention |
These are automatically flagged and deducted:
| Pattern | Deduction | Fix |
|---|---|---|
| Inter/Roboto/Arial font | -25 type | Use distinctive typeface |
| Purple gradient on white | -30 color | Choose bold, non-generic palette |
| Cards inside cards | -20 layout | Flatten hierarchy |
| No CSS custom properties | -15 color | Use variables for consistency |
| outline:none without replacement | -20 a11y | Add :focus-visible styles |
| No media queries | -15 layout | Add responsive breakpoints |
| Pure black text (#000) | -5 color | Tint dark colors |
| Off-grid spacing | -3 each layout | Snap to 4px grid |
The OPTIK landing page includes a live code analyzer. Paste any HTML or CSS, get an instant score with category breakdowns and specific issues.
git clone https://github.com/dragoon0x/optik.git
cd optik
node build.js # Compile skills to all providers
node test/test.js # Run test suite (514 checks)Apache 2.0. See LICENSE.
The core design intelligence skill builds on concepts from Anthropic's official frontend-design skill.
Built by dragoon0x
This project is experimental software, built out of curiosity and shared for educational purposes only.
This software is provided "AS IS" without warranty of any kind. The author makes no claims about the accuracy or reliability of design analysis, scoring, or recommendations. Design quality is subjective — scores are heuristic approximations, not definitive assessments. This is not a substitute for professional design review or accessibility auditing. Not affiliated with Anthropic, Cursor, Google, OpenAI, GitHub, Vercel, or any AI tool vendor.
DYOR — Do Your Own Research. See DISCLAIMER.md for full terms.