Figma MCP Complete
Exports Figma designs as ready-to-use HTML/CSS and validates pixel-perfect accuracy with image comparison and visual regression testing.
README
šØ Figma MCP Complete
By Nour Shamrok
Figma to Code MCP + Pixel Perfect Validation
Export Figma designs as ready-to-use HTML/CSS + Validate pixel-perfect accuracy āØ
š Table of Contents
- Overview
- Features
- Installation
- Quick Start
- Usage
- Examples
- API Reference
- Pixel Perfect Validation
- GitHub Actions CI/CD
- Troubleshooting
- Contributing
- License
šÆ Overview
Figma MCP Complete solves the biggest problem in design-to-code workflows:
The Problem ā
Figma MCP (existing):
ā Returns only JSON metadata
ā Claude guesses colors, fonts, spacing
ā Needs manual fixes
ā 30 minutes per component ā±ļø
Result: Frustration, mistakes, wasted time š«
The Solution ā
Figma MCP Better (this project):
ā Returns Screenshot + CSS + HTML + JSON
ā Claude knows exactly what to build
ā Ready-to-use code
ā 30 seconds per component ā”
Result: Perfect accuracy, zero iterations š
⨠Features
šØ Figma MCP Better
- ā Screenshot Export - Visual reference for accuracy
- ā CSS Generation - Production-ready styles
- ā HTML Code - Copy-paste ready components
- ā Design Tokens - Automated design system extraction
- ā High DPI Support - 2x resolution for Retina displays
ā Pixel Perfect Validator
- ā Image Comparison - Pixel-by-pixel accuracy checking
- ā Visual Regression Testing - Percy.io integration
- ā Heatmap Generation - See exact differences
- ā Responsive Testing - Mobile, Tablet, Desktop
- ā CI/CD Integration - GitHub Actions workflow
š Color & Typography Verification
- ā RGB to Hex conversion
- ā Font accuracy checking
- ā Spacing validation (padding, margin, gap)
- ā Shadow & effects verification
- ā Border radius matching
š Installation
Prerequisites
Step 1: Download & Extract
# Download figma-mcp-complete.zip
# Extract to your desired location
cd figma-mcp-complete
Step 2: Install Dependencies
# Install Node packages
npm install --save-dev playwright
# Install Python packages
pip install opencv-python numpy pillow
# Optional: For Percy.io validation
npm install --save-dev @percy/cli
Step 3: Setup Environment
Create .env file:
cat > .env << 'EOF'
FIGMA_TOKEN=figd_xxxxxxxxxxxxx
SIMULATOR_PORT=3000
PERCY_TOKEN=percy_xxxxxxxxxxxxx # Optional
EOF
Get your Figma Token: https://www.figma.com/developers/api#access-tokens
Step 4: Verify Installation
# Test the setup
node figma-mcp-comparison.js
# Output should show comparison of old vs new MCP ā
ā” Quick Start
Export Figma Design to HTML/CSS
# Set your Figma token
export FIGMA_TOKEN="figd_xxxxxxxxxxxxx"
# Export frame from Figma
node figma-mcp-better.js FILE_ID NODE_ID "Component Name"
# Results saved to:
# figma-export/
# āāā index.html ā Copy-paste this!
# āāā style.css ā Or this!
# āāā data.json ā Or everything here
# āāā tokens.json ā Design system
Get Figma IDs
- Open your Figma file:
https://www.figma.com/file/YOUR_FILE_ID/... - Copy the
FILE_IDfrom URL - Right-click on a frame ā Copy link ā Extract
NODE_ID
Example:
URL: https://www.figma.com/file/abc123def456/MyDesign?node-id=789:123
FILE_ID: abc123def456
NODE_ID: 789:123
Validate Design Accuracy
# Start your simulator/app
npm run start:simulator &
# Capture simulator screenshots
npm run capture:simulator
# Compare with Figma
npm run pixel-perfect
# Results:
# ā
Similarity: 99.8%
# ā
All colors match
# ā
Typography verified
# ā
Pixel Perfect!
š Usage
Basic Workflow
# 1. Export from Figma
export FIGMA_TOKEN="..."
node figma-mcp-better.js FILE NODE "Button"
# 2. Get HTML/CSS
cat figma-export/index.html
# <button style="background-color: #FF5722; ...">
# 3. Copy to your project
# Done! āØ
Advanced: Export Multiple Frames
// export-all.js
const BetterFigmaMCP = require('./figma-mcp-better');
const mcp = new BetterFigmaMCP(process.env.FIGMA_TOKEN);
const frames = [
{ id: '123:45', name: 'Login Button' },
{ id: '123:46', name: 'Signup Form' },
{ id: '123:47', name: 'Dashboard Header' }
];
(async () => {
for (const frame of frames) {
const result = await mcp.exportFrame(FILE_ID, frame.id, frame.name);
await mcp.saveExport(result, `./exports/${frame.name}`);
console.log(`ā
Exported: ${frame.name}`);
}
})();
Run:
node export-all.js
Integration with Claude AI
// Use with Claude API
const BetterFigmaMCP = require('./figma-mcp-better');
const figmaData = await mcp.exportFrame(fileId, nodeId);
// Claude now gets:
// - figmaData.screenshot ā Visual reference
// - figmaData.css ā Styling
// - figmaData.html ā Code
// - figmaData.designTokens ā System
// Result: Claude writes perfect code! šÆ
š Examples
Example 1: Button Component
export FIGMA_TOKEN="figd_xxx"
node figma-mcp-better.js abc123 "456:789" "Primary Button"
Output: figma-export/index.html
<!DOCTYPE html>
<html>
<head>
<meta charset="UTF-8">
<title>Primary Button</title>
<style>
.container {
background-color: #FF5722;
color: #FFFFFF;
font-size: 16px;
font-family: Inter;
font-weight: 700;
padding: 12px 24px;
border-radius: 8px;
box-shadow: 4px 4px 8px rgba(0,0,0,0.3);
}
</style>
</head>
<body>
<button class="container">Click me</button>
</body>
</html>
Example 2: Design System Extraction
node figma-mcp-better.js abc123 "def:456" "Design System"
Output: figma-export/tokens.json
{
"colors": {
"primary": "#FF5722",
"secondary": "#2196F3",
"text": "#212121"
},
"typography": {
"heading": {
"fontFamily": "Montserrat",
"fontSize": 32,
"fontWeight": 700,
"lineHeight": 1.2
},
"body": {
"fontFamily": "Open Sans",
"fontSize": 16,
"fontWeight": 400,
"lineHeight": 1.5
}
},
"spacing": {
"xs": "4px",
"sm": "8px",
"md": "16px",
"lg": "24px"
}
}
Example 3: Pixel Perfect Validation
npm run pixel-perfect
Output:
šØ Comparing 5 screens...
1. login-mobile: 99.8% ā
2. dashboard-mobile: 99.5% ā
3. profile-mobile: 99.9% ā
4. settings-mobile: 99.2% ā
5. onboarding-mobile: 99.7% ā
============================================================
PIXEL PERFECT VALIDATION RESULTS
============================================================
Average Similarity: 99.62%
Threshold: 99.0%
Screens Checked: 5
ā
ALL SCREENS PASSED PIXEL PERFECT VALIDATION!
š API Reference
BetterFigmaMCP Class
Constructor
const mcp = new BetterFigmaMCP(figmaToken);
Methods
exportFrame(fileId, nodeId, frameName)
Export a single frame.
const result = await mcp.exportFrame(
'abc123def456',
'789:123',
'Login Button'
);
// Returns:
{
name: 'Login Button',
screenshot: 'https://...',
css: { ... },
html: '<button>...</button>',
designTokens: { ... },
json: { ... }
}
saveExport(data, outputDir)
Save export to files.
await mcp.saveExport(result, './exports');
// Creates:
// ./exports/index.html
// ./exports/style.css
// ./exports/data.json
// ./exports/tokens.json
extractCSS(node)
Extract CSS from Figma node.
const css = mcp.extractCSS(figmaNode);
// Returns: { backgroundColor, fontSize, ... }
extractDesignTokens(file)
Extract design system tokens.
const tokens = mcp.extractDesignTokens(figmaFile);
// Returns: { colors, typography, spacing, shadows }
ā Pixel Perfect Validation
How It Works
1. Export Figma screenshot ā figma-baseline/
2. Capture simulator screenshot ā simulator-screenshots/
3. Compare pixel-by-pixel ā reports/
4. Generate heatmap of differences
Run Validation
# Capture simulator
npm run capture:simulator
# Compare with Figma
npm run pixel-check
# Or full validation
npm run pixel-perfect
Thresholds
| Similarity | Status | Action |
|---|---|---|
| 99%+ | ā PASS | Merge! |
| 95-99% | ā ļø REVIEW | Check heatmap |
| <95% | ā FAIL | Fix design |
View Heatmap
Differences appear in reports/heatmap_*.png:
- š“ Red = Large differences
- š” Yellow = Small differences
- šµ Blue/Green = Identical
š GitHub Actions CI/CD
Automatic Validation on Push
The project includes GitHub Actions workflow that:
- ā Validates on every push
- ā Checks pixel-perfect accuracy
- ā Blocks merge if validation fails
- ā Comments on PRs with results
Setup GitHub Actions
# Copy workflow file
.github/workflows/pixel-perfect.yml
# Commit to GitHub
git add .github/
git commit -m "Add pixel perfect CI/CD"
git push
# Set branch protection
GitHub ā Settings ā Branches
ā Require status checks to pass
ā Select "Pixel Perfect Validation"
Require Pixel Perfect
Now you cannot merge without passing validation! š
PR created
ā
GitHub Actions runs
ā
Pixel Perfect test passes?
ā
ā No ā PR blocked š«
ā
ā
Yes ā Can merge ā
š Troubleshooting
Installation Issues
"npm: command not found"
# Install Node.js
https://nodejs.org/
# Verify:
node --version
npm --version
"python3: command not found"
# Install Python
https://python.org/downloads
# Verify:
python3 --version
"ModuleNotFoundError: No module named 'cv2'"
# Reinstall Python packages
pip install --upgrade opencv-python numpy pillow
Runtime Issues
"FIGMA_TOKEN not set"
# Set token
export FIGMA_TOKEN="figd_xxxxxxxxxxxxx"
# Or create .env file
echo "FIGMA_TOKEN=figd_xxxxxxxxxxxxx" > .env
"Node FILE_ID NODE_ID not found"
# Get correct IDs from Figma URL
# https://www.figma.com/file/YOUR_FILE_ID/...?node-id=YOUR_NODE_ID
# Verify they're correct
echo $FIGMA_TOKEN
"Screenshots don't match"
# Check DPI consistency
# Ensure both screenshots are 2x scale (Retina)
# Manually verify
open figma-export/index.html
open reports/heatmap_*.png
Performance
Slow export?
# Reduce resolution
node figma-mcp-better.js FILE NODE --scale=1
Memory issues?
# Process one frame at a time
# Or increase Node memory
node --max-old-space-size=4096 figma-mcp-better.js
š Comparison: Old vs New
| Feature | Old Figma MCP | Figma MCP Complete |
|---|---|---|
| Screenshot | ā | ā |
| CSS | ā | ā |
| HTML | ā | ā |
| Design Tokens | ā | ā |
| Color Format | RGB(0-1) š© | #HEX š |
| Font Size | Just number | px (web standard) |
| Ready to Use | ā 30 min | ā 30 sec |
| Claude Accuracy | 60% | 99%+ |
| Iterations Needed | Many | Zero |
š¤ Contributing
Contributions welcome!
# Fork & clone
git clone https://github.com/YOUR_FORK/figma-mcp-complete.git
cd figma-mcp-complete
# Create branch
git checkout -b feature/amazing-feature
# Make changes
# Test thoroughly
npm test
# Push & create PR
git push origin feature/amazing-feature
Development
# Run tests
npm test
# Lint code
npm run lint
# Build docs
npm run docs
š License
MIT License - See LICENSE file
š Acknowledgments
- Built with ā¤ļø for designers and developers
- Inspired by real design-to-code workflow problems
- Thanks to Figma & Anthropic communities
š Support
Documentation
- š Full Guide
- š Pixel Perfect Guide
- ā” Quick Start
Issues & Questions
- š GitHub Issues
- š¬ Discussions
- š§ Email: shamroknour057@gmail.com
Community
- š¤ Claude Community
- šØ Figma Community
- š» GitHub Discussions
ā If you find this useful, please star! ā
Star this repo to show support āØ
Share with your team š„
Contribute improvements š
šÆ Roadmap
- [ ] Web UI for exports
- [ ] Figma plugin integration
- [ ] Multiple design system formats
- [ ] Real-time validation
- [ ] Browser extension
- [ ] Cloud storage integration
š Stats
ā
Lines of code: 5000+
ā
Supported Figma nodes: 15+
ā
Export formats: 5+ (HTML, CSS, JSON, etc)
ā
Validation accuracy: 99%+
ā
Performance: <30s per component
Made with ā¤ļø for better design-to-code workflows
Created by Nour Shamrok š
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.