Provides access to Figma files and prototypes for AI coding agents via an MCP server.
Configuration
View docs{
"mcpServers": {
"jayzeedesign-figma-mcp": {
"command": "figma-mcp",
"args": [
"--figma-api-key=your_figma_key"
]
}
}
}You enable a MCP server that lets your AI agents interact with Figma files and prototypes directly. This server runs locally or alongside your editor, so you can access Figma data within your development workflow and automate tasks that involve designs and prototypes.
To use the MCP server, run the local server command from your project and provide your Figma API key when prompted. Then connect your MCP client (such as your editor’s built‑in MCP client) to the local server to request access to Figma files and prototypes as part of your coding workflows.
Prerequisites you need before installing the MCP server are Python and a shell with curl available on your system. You will also use uv to manage a Python virtual environment.
Step by step commands to set up and run the MCP server locally:
pipx install figma-mcpcurl -LsSf https://astral.sh/uv/install.sh | sh
uv venv
source .venv/bin/activate
uv syncStart the MCP server locally to expose the Figma access point via the standard Python package entry. You can test running the module directly.
python -m figma_mcp.mainConfiguration options shown in examples include per-project MCP server entries for editors and IDEs. You can tailor how you add the MCP server to your editor, Cursor, Windsurf, or other IDEs by supplying the appropriate command and arguments.
The MCP server for Figma is configured to run the figma-mcp command with your API key passed as an argument. You’ll typically add this to your project’s MCP configuration so your editor can reach the server when needed.
{
"mcpServers": {
"figma_python": {
"command": "figma-mcp",
"args": [
"--figma-api-key=your_figma_key"
]
}
}
}