CustomFigmaMcp
Analyzes Figma design files through natural language queries, enabling component detection, color extraction, and consistency checks via AI assistants like Claude in Cursor.
README
šØ Figma MCP - My First Custom MCP Project Hey! This is my personal experiment with building a custom MCP (Model Context Protocol) server that connects Figma to AI assistants like Claude in Cursor. It's my first time creating an MCP, so bear with me - but it actually works pretty well for analyzing Figma design files! What does this thing do? Basically, I got tired of manually checking my Figma designs for consistency issues, so I built this tool that lets me ask Claude (through Cursor) questions about my designs in plain English. š Design File Analysis
Get a quick overview of your entire Figma file - pages, colors, styles, the whole thing Dig into specific components or groups when you need details Check if your design system is being used consistently (spoiler: it probably isn't)
š§© UI Component Detective Work
Automatically finds buttons, cards, input fields, etc. in your designs Counts how many times you've used each component (useful for cleaning up) Spots inconsistencies in spacing, colors, and sizes that you might have missed
š Reports That Actually Help
Shows which components you use most/least (goodbye, forgotten components) Extracts all the colors you're using (yes, even that random blue you used once) Typography audit - finds all your text styles and font chaos Complete inventory of your design assets
š¤ The Cool AI Part
Ask questions about your designs in normal English Get suggestions for improvements (when it works, it's pretty neat) No more manually counting components or checking consistency
Getting This Thing Running What You Need
Node.js 18+ A Figma Personal Access Token (I'll show you how to get this) Cursor IDE
Setup Steps
Get your Figma token:
Go to Figma Settings ā Personal Access Tokens Click "Create new token" Give it a name like "My MCP experiment" Copy the token (it starts with figd_ - don't lose it!) Also copy your Figma file URL
Set up Cursor MCP:
Find your Cursor MCP settings file:
Mac: ~/.cursor-server/mcp_settings.json Windows: %APPDATA%/Cursor/mcp_settings.json Linux: ~/.config/cursor/mcp_settings.json
Add this (replace the paths and tokens with your actual ones): json{ "mcpServers": { "figma-mcp": { "command": "node", "args": ["/full/path/to/this/project/dist/index.js"], "env": { "FIGMA_ACCESS_TOKEN": "figd_your_actual_token_here", "FIGMA_PROJECT_URL": "https://www.figma.com/file/your-file-id/your-file-name" } } } }
Turn on MCP in Cursor:
Open Cursor settings (Cmd/Ctrl + ,) Find Features ā Model Context Protocol Turn on MCP Support Restart Cursor (important!)
How to Use It Once it's working, you can chat with Claude about your designs: Ask Simple Questions @figma-mcp What's in my Figma file? Check Your Components @figma-mcp How many different button styles do I have? Find Inconsistencies @figma-mcp Are my colors consistent across pages? Get Component Stats @figma-mcp Which components am I not using much? Deep Dive into Specifics @figma-mcp Show me all the cards in my design and their properties Things I've Learned (aka Troubleshooting) If it's not working:
Path issues: Use the full, absolute path to your built project Token problems: Make sure your token starts with figd_ and has access to your file Cursor restart: Seriously, restart Cursor after changing MCP settings Build the project: Run npm run build first if you're using the source code
Common "oops" moments:
Forgetting to restart Cursor (I did this like 5 times) Using a relative path instead of absolute path Token doesn't have permission to access the Figma file
What I Built This With
TypeScript (because I like making my life complicated) Figma API (surprisingly well-documented) MCP SDK (still learning this) A lot of trial and error
Honest Assessment What works well:
Basic file analysis is solid Component detection is pretty accurate Color extraction works great Integration with Cursor feels natural
What could be better:
Error handling could be more robust Some complex nested components confuse it I haven't tested it on huge files yet
What I want to add next:
Better component classification Export capabilities Maybe some design suggestions
If You Want to Help This is a learning project, so if you spot issues or have ideas:
Open an issue if something breaks PRs welcome if you want to improve it Or just tell me what you think!
Personal note: This started as a weekend project because I was manually checking design consistency and thought "there has to be a better way." Turns out building your own MCP is actually pretty fun, and now I can ask Claude about my designs instead of clicking through every page. Not bad for a first attempt! Made with ā and curiosity about what AI can do with design files.
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.