Skill: frontend-design-scanner

Activation Contract

Activate this skill when the user needs to:

• Scan a frontend codebase to understand its design system and CSS architecture
• Detect the presence and version of Tailwind CSS, CSS frameworks, or preprocessors
• Identify the design methodology: Atomic Design, BEM, SMACSS, ITCSS, OOCSS, or utility-first
• Extract branding tokens: fonts, sizes, weights, border radius, colors, spacing scales
• Map the component hierarchy and identify base components (buttons, inputs, selects, textareas)
• Audit the consistency of design tokens across the codebase

Do not activate for pure backend projects or when only trivial CSS is present.

Hard Rules

1. Always use the detection scripts in assets/ when scanning; do not rely solely on file inspection.
2. Never assume the architecture; detect it from class naming patterns, folder structure, and config files.
3. Report confidence levels for each detection (High / Medium / Low).
4. Prefer AST parsing over regex when analyzing CSS/SCSS/JSX/Vue/Svelte files.
5. Normalize all extracted tokens into a standard JSON schema defined in assets/token-schema.json.
6. Respect .gitignore and skip node_modules, .next, dist, build folders.

Decision Gates

Execution Steps

Phase 1: Environment Detection (5 steps)

1. Detect CSS Frameworks & Tools

- Check package.json for: tailwindcss, postcss, sass, less, styled-components, @emotion/, bootstrap, bulma, foundation - Check for config files: tailwind.config.js|ts, postcss.config.js, vite.config., webpack.config.* - Record versions from package.json or lock files - Report: framework name, version, config location

1. Detect Tailwind CSS Specifics (Ultra-Precise)

- Check package.json for tailwindcss version (v2, v3, v4) - Detect v4: CSS-based config with @import "tailwindcss" and @theme directives - Detect v3: tailwind.config.js|ts|mjs|cjs with theme.extend - Extract theme variables: - v4: Parse CSS for --color-, --font-, --spacing-, --radius-, --shadow-, --breakpoint- - v3: Parse JS config for theme.extend.colors, fontFamily, spacing, borderRadius - Detect plugins: @tailwindcss/forms, @tailwindcss/typography, tailwindcss-animate - Classify customization level: default | light | moderate | heavy - Report: version, config type (js-config vs css-config), theme summary, customizations

1. Detect CSS Architecture Methodology

- Scan all .css, .scss, .less, .sass, .styl files - Look for naming patterns: - BEM: .block__element--modifier pattern - Atomic Design: folders named atoms/, molecules/, organisms/, templates/, pages/ - SMACSS: folders named base/, layout/, module/, state/, theme/ - ITCSS: folders named settings/, tools/, generic/, elements/, objects/, components/, trumps/ - OOCSS: separation of structure and skin patterns - Utility-first: prevalence of single-purpose classes (e.g., flex, pt-4, text-center) - Calculate methodology confidence score based on: - Folder structure matches (40%) - Class naming patterns (40%) - File organization (20%) - Report: primary methodology, secondary methodology (if any), confidence scores

1. Detect Component Architecture

- Scan component files (.jsx, .tsx, .vue, .svelte, .astro) - Build dependency graph (imports/exports) - Identify atomic levels if Atomic Design is detected: - Atoms: basic HTML wrappers (Button, Input, Label) - Molecules: composed atoms (SearchBar, FormField) - Organisms: complex sections (Header, Hero, Footer) - Templates: page layouts - Pages: route components - Report: component tree depth, component count by level, most reused components

1. Detect Base Components

- Search for component patterns matching: - Buttons: Button, Btn, button, ActionButton - Inputs: Input, TextField, TextInput, FormInput - Selects: Select, Dropdown, SelectField - Textareas: Textarea, TextArea, TextField (multiline) - Labels: Label, FormLabel - Checkboxes/Radios: Checkbox, Radio, Switch, Toggle - Extract their props interfaces to detect variant systems - Report: component locations, variant systems, prop signatures

Phase 2: Token Extraction (4 steps)

1. Extract Font Tokens

- Parse CSS/SCSS for font-family, @font-face, CSS variables like --font-* - Check tailwind.config.js theme.fontFamily - Extract: font families, fallback stacks, font weights used (400, 500, 700), font styles (italic, normal) - Report: primary font, secondary font, monospace font, all weights, all sizes

1. Extract Font Size Tokens

- Parse CSS for font-size declarations - Check Tailwind config theme.fontSize or CSS vars --text-, --font-size- - Normalize to pixel or rem values - Report: all sizes sorted ascending, base size, scale ratio

1. Extract Border Radius Tokens

- Parse CSS for border-radius declarations - Check Tailwind config theme.borderRadius or CSS vars --radius-, --rounded- - Report: all radius values, common patterns (0, 4px, 8px, 9999px/pill)

1. Extract Color Tokens

- Parse CSS for color declarations: color, background-color, border-color - Check CSS variables: --color-, --primary, --secondary, --bg- - Check Tailwind config theme.colors, theme.extend.colors - Report: primary, secondary, accent, neutrals, semantic colors (success, error, warning), dark mode colors

Phase 3: Analysis & Reporting (3 steps)

1. Analyze Consistency

- Check if border radius values are consistent with component types - Verify font sizes follow a modular scale - Detect hardcoded values vs token-based values - Report: consistency score, hardcoded value locations, token coverage percentage

1. Generate Token Map

- Output standardized JSON following assets/token-schema.json - Include metadata: framework, architecture, confidence, scan timestamp - Group tokens by category: colors, typography, spacing, borders, shadows

1. Summarize & Recommend

- Generate executive summary: 3-5 bullet points of key findings - Flag inconsistencies or tech debt - Recommend next steps: token consolidation, migration to CSS vars, Tailwind config cleanup, etc.

Output Contract

Return a structured report containing:

{
  "scan_summary": {
    "project_path": "string",
    "frameworks": [{"name": "string", "version": "string", "config_path": "string"}],
    "css_architecture": {
      "primary": "string",
      "secondary": "string|null",
      "confidence": "number"
    },
    "component_stats": {
      "total": "number",
      "by_level": {"atoms": "number", "molecules": "number", "organisms": "number"},
      "base_components": {
        "buttons": ["string"],
        "inputs": ["string"],
        "selects": ["string"],
        "textareas": ["string"]
      }
    }
  },
  "tokens": {
    "fonts": {
      "families": [{"name": "string", "fallback": "string", "source": "string"}],
      "sizes": [{"value": "string", "px": "number", "usage_count": "number"}],
      "weights": ["number"],
      "styles": ["string"]
    },
    "border_radius": [{"value": "string", "px": "number", "usage_count": "number"}],
    "colors": {
      "primary": "string",
      "secondary": "string",
      "accent": "string",
      "neutrals": ["string"],
      "semantic": {"success": "string", "error": "string", "warning": "string"}
    }
  },
  "analysis": {
    "consistency_score": "number",
    "hardcoded_values": [{"file": "string", "line": "number", "value": "string"}],
    "token_coverage": "number",
    "recommendations": ["string"]
  }
}

Also return:

• Files created or modified
• Detection confidence levels for each section
• Any errors or skipped files

Enhanced Detection Features (v2.0)

1. Framework UI Detection

Automatically detects popular Tailwind-based component libraries:

• shadcn/ui - Via class-variance-authority or components/ui folder
• Radix UI - @radix-ui/react-* packages
• DaisyUI - daisyui package
• Flowbite - flowbite and flowbite-react
• Headless UI - @headlessui/react (official)
• Tailwind UI - @tailwindcss/ui
• Animation libraries - framer-motion, tailwindcss-animate

2. Real Usage Analysis

Scans source files to detect:

• All Tailwind classes used with frequency count
• Arbitrary values - w-[123px], bg-[#1da1f2]
• Responsive prefixes - sm:, md:, lg: usage stats
• Dark mode classes - dark: prefix usage
• Hardcoded values - Colors outside of theme
• Token coverage - Percentage of custom tokens being used
• Top 20 most used classes

3. Base Component Detection

Identifies common UI components by filename and content:

• Buttons - button, btn, action-button
• Inputs - input, text-field, form-input
• Selects - select, dropdown, autocomplete
• Textareas - textarea, text-area
• Cards - card, info-card
• Modals - modal, dialog, overlay
• Tables - table, data-table
• Navigation - nav, navbar, sidebar
• Tabs - tabs, tab-list
• Alerts - alert, toast, notification
• Badges - badge, tag, pill
• Avatars - avatar, user-avatar
• Tooltips - tooltip, popover

FAANG & Big Tech Patterns

Based on research of major tech companies' design systems:

Key Learnings from FAANG

1. Atomic CSS Dominance: All major companies use or are migrating to atomic/utility CSS (Meta's StyleX reduced CSS by 80%)
2. Build-Time Compilation: Styles are co-located with components but compiled/transformed at build time
3. Design Tokens Standardization: JSON-first token definitions (Amazon Style Dictionary) or CSS variables (Meta, Google)
4. Theme Support: CSS custom properties for runtime theming (dark mode, brand switching)
5. Component Co-location: Styles live next to components, never in separate global CSS files
6. Confidence Scoring: Use folder structure (40%), naming patterns (40%), file organization (20%) for detection

Enhanced Detection Priority

1. Meta/StyleX: Detect @stylexjs/stylex, stylex.create(), stylex.defineVars()
2. Google/Material: Detect @material/, .mdc- classes, --mdc-theme-* variables
3. Amazon/Tokens: Detect style-dictionary dependency, tokens/ folder, config.json
4. Tailwind: Detect tailwind.config.*, @tailwind directives, utility classes
5. BEM/Atomic: Detect folder structure and naming patterns
6. CSS-in-JS: Detect styled-components, @emotion/*, linaria

References

• assets/detect-tailwind.js — Tailwind CSS detection and parsing script
• assets/detect-css-architecture.js — CSS methodology detection (BEM, Atomic, SMACSS, ITCSS)
• assets/detect-component-tree.js — Component hierarchy and base component detection
• assets/extract-tokens.js — Token extraction from CSS/SCSS/JSX/Vue
• assets/detect-css-in-js.js — CSS-in-JS library detection (styled-components, Emotion, StyleX)
• assets/token-schema.json — Standardized token output schema
• assets/scan-directory-tree.js — Directory tree scanner (paths, .gitignore aware, CLI + API)
• references/architectures.md — CSS architecture patterns reference
• references/tailwind-detection.md — Tailwind-specific detection patterns
• references/faang-patterns.md — FAANG design system patterns and detection algorithms