MCP server

figma-ui-mcp: Draw and Read Figma with an AI Agent

figma-ui-mcp lets an AI coding agent draw UI on the Figma canvas through JavaScript and read existing designs back as structured data, with no Figma API key.

Run
claude mcp add figma-ui-mcp -- npx figma-ui-mcp
View on GitHub

Stars, license and last push from GitHub. Install count as reported by the open skills directory.

What is figma-ui-mcp?

figma-ui-mcp is a two-way MCP bridge for Figma. An AI coding agent writes frames, shapes and text onto the canvas through a Figma Desktop plugin, and reads node trees, colors, typography and CSS back out. It needs no Figma API key.

It is a bidirectional Figma bridge for Model Context Protocol clients. The agent can write to Figma, drawing frames, shapes, text and prototypes, and read from it, pulling node trees, colors and typography. A plugin inside Figma Desktop does the actual work.

How the bridge works

The MCP server runs a small HTTP server on localhost port 38451. The Figma plugin long-polls it for work, runs the operations against the open document, and sends results back. Because everything stays on localhost, no Figma API key is needed.

Figma Desktop is required. Figma's web app does not allow localhost network access, so the plugin cannot talk to the server from a browser tab.

What you can do with it

  • Draw a screen from a prompt, then refine it in Figma.
  • Read a design as AI-ready context, including flex layout, token-resolved colors and component variants.
  • Export ready-to-paste CSS for a node.
  • Generate a rule sheet of tokens, typography and components for your project.

Best for

  • Designers and developers who want an agent working directly on a Figma file.
  • Teams moving a finished design toward code. If you prefer a different bridge, compare Figma Unified MCP and Figma Relai MCP.

Use it with UiChemy

After the design is settled in Figma, UiChemy's Figma to WordPress converts the frames into editable Elementor or Gutenberg pages. Get Started at uichemy.com.