Figma Design Tokens Generator

Overview

This skill enables extraction of design tokens (colors, typography, spacing, etc.) from Figma designs through the Figma MCP server and transforms them into production-ready format files. Generate CSS custom properties, SCSS variables, JSON, W3C Design Tokens Format, TypeScript types, and documentation from Figma variables.

Prerequisites

Required:

• Figma MCP server installed and configured in ~/.claude/config.json
• Figma file with design variables (or elements with styles for fallback extraction)
• View or edit access to the Figma file

Setup verification:

• If user reports MCP connection issues, consult references/troubleshooting.md
• For installation help, direct user to README.md

Workflow Decision Tree

Choose the appropriate workflow based on the user's goal:

A. Selection-Based Extraction

• User has specific frames/components to extract
• Working with subset of design system
• Quick extraction for specific use case

B. Full Design System Extraction

• User wants all tokens from entire file
• Building comprehensive design system
• Multiple themes/modes to handle

Selection-Based Extraction Workflow

Step 1: Extract Variables from Figma

Use the get_variable_defs MCP tool to extract Figma variables:

Use the Figma MCP get_variable_defs tool with:
- file_url: <figma-file-url>
- node_ids: <comma-separated-node-ids> (optional, extracts all if omitted)

Returns: JSON with variable definitions including names, values, types, modes, and references.

Fallback: If no variables exist, the extraction script automatically extracts tokens from node styles (fills, strokes, typography, spacing) with duplicate detection.

Step 2: Apply Core Token Generation Workflow

Proceed to "Core Token Generation Workflow" section below.

Full Design System Workflow

Step 1: Get File Metadata (Optional)

Use get_metadata MCP tool to understand file structure:

Use the Figma MCP get_metadata tool with:
- file_url: <figma-file-url>

This returns layer IDs, names, types, and hierarchy.

Step 2: Extract All Variables

Use get_variable_defs without node_ids:

Use the Figma MCP get_variable_defs tool with:
- file_url: <figma-file-url>

Step 3: Apply Core Token Generation Workflow

Proceed to "Core Token Generation Workflow" section below.

Core Token Generation Workflow

This workflow applies to both extraction approaches above.

Step 1: Transform to W3C DTCG Format

Use the bundled extract_tokens.py script to transform Figma data:

python figma-design-tokens/scripts/extract_tokens.py \
  --figma-data <path-to-mcp-output.json> \
  --output-path <output-tokens.json> \
  --token-types colors,typography,spacing

Key Transformations:

• Figma COLOR → W3C color tokens (sRGB color space)
• Figma FLOAT → W3C dimension or number tokens
• Figma STRING → W3C fontFamily or custom tokens
• Variable references → W3C alias syntax {token.name}
• Variable modes → Organized token groups or separate files

Token Types:

• colors - Color variables
• typography - Font families, sizes, weights, line heights
• spacing - Spacing, margins, padding, dimensions
• all - All available token types (default)

Step 2: Generate Output Formats

Use the bundled transform_tokens.py script:

python figma-design-tokens/scripts/transform_tokens.py \
  --input <tokens.json> \
  --format css,scss,json,typescript,documentation \
  --naming-convention kebab-case \
  --output-dir <output-directory>

Available Formats:

1. CSS Custom Properties - :root { --color-primary: #0066cc; }
2. SCSS Variables - $color-primary: #0066cc;
3. JSON - Style Dictionary compatible
4. TypeScript - Type-safe definitions with as const
5. Documentation - Markdown tables and usage examples

Naming Conventions:

• kebab-case (default) - color-primary-500
• camelCase - colorPrimary500
• snake_case - color_primary_500
• bem - color__primary--500

Step 3: Validate Output

Use the bundled validate_tokens.py script:

python figma-design-tokens/scripts/validate_tokens.py \
  --input <output-tokens.json> \
  --report <validation-report.md>

Validation Checks:

• W3C DTCG specification compliance
• Required properties presence ($value)
• Token type consistency
• Circular reference detection
• Value format correctness

Step 4: Organize Files (Optional)

For category-specific files:

python figma-design-tokens/scripts/transform_tokens.py \
  --input <tokens.json> \
  --format css,scss,json \
  --organize-by-category \
  --output-dir <output-directory>

Generated structure:

output/
├── colors.tokens.json
├── colors.css
├── colors.scss
├── spacing.tokens.json
├── spacing.css
├── spacing.scss
├── typography.tokens.json
├── typography.css
└── typography.scss

Multi-Theme Handling

If Figma file contains variable modes (light/dark themes):

Option A: Separate Files Per Theme

# Extract each mode separately
python figma-design-tokens/scripts/extract_tokens.py \
  --figma-data <mcp-output.json> \
  --mode light \
  --output-path <tokens-light.json>

python figma-design-tokens/scripts/extract_tokens.py \
  --figma-data <mcp-output.json> \
  --mode dark \
  --output-path <tokens-dark.json>

Option B: Token References

Generate base tokens with mode-specific tokens referencing base:

{
  "color": {
    "blue-500": {
      "$type": "color",
      "$value": "rgb(0, 102, 204)"
    },
    "primary": {
      "$value": "{color.blue-500}"
    }
  }
}

Then create theme files that override references.

Token Naming Conventions

Default (kebab-case)

color-primary-500
spacing-medium
font-family-body

Custom Prefixes

Use --category-prefix to add category prefixes:

--category-prefix color:clr,spacing:sp,typography:type

Results in: clr-primary-500, sp-medium, type-body

Semantic Name Standardization (NEW)

Automatically standardize Figma token names to industry-standard semantic conventions:

python figma-design-tokens/scripts/extract_tokens.py \
  --figma-data <mcp-output.json> \
  --output-path <tokens.json> \
  --standardize-names

What It Does:

1. Color Name Standardization

   • Maps Figma color names to semantic standards
   • Brand Blue → primary
   • Red or Danger → error
   • Green or Positive → success
   • Yellow or Alert → warning
   • Supports: primary, secondary, tertiary, success, warning, error, info, neutral, background, foreground, accent, link

2. Size Standardization (T-shirt Sizing)

   • Normalizes dimensions to t-shirt scale: xs, sm, md, lg, xl, 2xl, 3xl, 4xl
   • Applies to spacing, width, height, border-radius, font-size
   • Name-based: Small → sm, Extra Large → xl
   • Value-based: 4px → xs, 16px → md, 64px → 2xl

Custom Mappings:

Create a JSON file with custom semantic mappings:

{
  "colors": {
    "brand": ["brand", "company", "corporate"],
    "disabled": ["disabled", "inactive", "muted"]
  },
  "sizes": {
    "tiny": ["tiny", "micro", "mini"],
    "huge": ["huge", "jumbo", "massive"]
  }
}

Use with --name-mappings:

python figma-design-tokens/scripts/extract_tokens.py \
  --figma-data <mcp-output.json> \
  --output-path <tokens.json> \
  --standardize-names \
  --name-mappings <custom-mappings.json>

Benefits:

• Consistent naming across projects
• Easier onboarding for developers
• Better integration with component libraries
• Semantic clarity (intent over appearance)

Token Organization Patterns

Primitive vs Semantic Tokens

Primitive Tokens (base values):

{
  "blue": {
    "500": { "$value": "#0066cc" }
  }
}

Semantic Tokens (context-specific):

{
  "color": {
    "primary": { "$value": "{blue.500}" }
  }
}

The scripts automatically identify primitive vs semantic based on variable references.

File Organization Strategies

Use --organization-strategy or --organize-by-category flag:

• Single File - All tokens in one tokens.json (simple projects)
• Category Files - Separate files per category (medium projects)
• Primitive + Semantic Split - Advanced organization (large design systems)

Script Usage Reference

extract_tokens.py

Purpose: Transform Figma MCP data to W3C DTCG format

Key Arguments:

• --figma-data - Path to Figma MCP JSON output (required)
• --output-path - Output file path (required)
• --token-types - Comma-separated types: colors, typography, spacing, all
• --mode - Figma variable mode (e.g., light, dark)
• --pretty - Pretty-print JSON output
• --standardize-names - Enable semantic name standardization (optional)
• --name-mappings - Path to custom mappings JSON file (optional)

Help: python extract_tokens.py --help

transform_tokens.py

Purpose: Generate CSS, SCSS, JSON, TypeScript, documentation from W3C tokens

Key Arguments:

• --input - W3C DTCG JSON file (required)
• --format - Output formats: css, scss, json, typescript, documentation
• --naming-convention - Token naming style
• --output-dir - Output directory (required)
• --organize-by-category - Generate separate files per category

Help: python transform_tokens.py --help

validate_tokens.py

Purpose: Validate W3C DTCG compliance

Key Arguments:

• --input - W3C DTCG JSON file (required)
• --report - Validation report path (markdown)
• --strict - Treat warnings as errors

Help: python validate_tokens.py --help

Common Use Cases

Extract Brand Colors

1. Select brand color frames in Figma
2. Extract using get_variable_defs
3. Transform to CSS custom properties
4. Import into global stylesheet

Build Multi-Theme System

1. Define light/dark modes in Figma variables
2. Extract both modes separately
3. Generate theme-specific CSS files
4. Switch themes with CSS class toggling

Create Component Library Tokens

1. Extract component-specific spacing/sizing
2. Generate TypeScript definitions
3. Import into React/Vue components
4. Use type-safe token access

Generate Design System Documentation

1. Extract all design tokens
2. Generate documentation markdown
3. Include in design system site
4. Maintain single source of truth

Extract from Files Without Variables

1. Open Figma file (only has styles, no variables)
2. Select frames/components with design elements
3. Use extraction script (auto-falls back to node extraction)
4. Script extracts colors, typography, spacing from properties
5. Duplicates automatically skipped

Bundled Resources

Scripts

• scripts/extract_tokens.py - Transform Figma → W3C DTCG
• scripts/transform_tokens.py - W3C → Multiple formats
• scripts/validate_tokens.py - Validate W3C compliance

All scripts include --help for detailed usage.

References

• references/w3c-dtcg-spec.md - Complete W3C Design Tokens specification
• references/figma-mcp-tools.md - Figma MCP server tools documentation
• references/token-naming-conventions.md - Naming patterns and best practices
• references/troubleshooting.md - Error handling and solutions
• references/integration-examples.md - Build tool integration (Style Dictionary, PostCSS, React, Vue, etc.)
• references/advanced-usage.md - Custom types, batch processing, automation

Load these references when detailed information is needed.

Templates

• templates/w3c-tokens.template.json - W3C DTCG structure template
• templates/css-variables.template.css - CSS custom properties template
• templates/scss-variables.template.scss - SCSS variables template
• templates/typescript-types.template.ts - TypeScript definitions template
• templates/documentation.template.md - Documentation generation template

Quick Decision Guide

User says: "Extract design tokens from Figma" → Ask: Full file or specific selection? → Guide to appropriate workflow

User says: "No variables found" → Explain: Automatic fallback extracts from node styles → Verify: File has accessible elements with fills/text/spacing

User reports: "MCP server error" → Direct to: references/troubleshooting.md

User asks: "How to integrate with [tool]" → Direct to: references/integration-examples.md

User needs: Advanced features → Direct to: references/advanced-usage.md

Expected Output

When using this skill, deliver:

1. W3C DTCG JSON File - Specification-compliant design tokens
2. CSS Custom Properties - Ready for browser usage
3. SCSS Variables - For Sass-based projects
4. TypeScript Definitions - Type-safe token access
5. Documentation - Auto-generated reference guide
6. Validation Report - Compliance and quality checks

All files are production-ready and can be integrated directly into projects.

For detailed information, consult the bundled reference documentation in the references/ directory.