LucasGorgal/localfig

Overview

Turns the Figma desktop app into a local MCP server, exposing the full Figma Plugin API to AI agents. It reads and writes the open file, including tokens, components, exports, and undo, without a cloud API, token, or quota. One command registers it with Claude Code, Cursor, VS Code, Windsurf, Cline, Gemini CLI, Codex, and other MCP clients.

Best for

Best for
Developers who want to give AI coding agents direct, local control over their active Figma files.

Use cases

• Read and write design tokens in the open Figma file
• Create or modify components programmatically from an AI agent
• Export assets and trigger undo operations via the Plugin API

How to use

Install

npx -y localfig --setup

Tools exposed

• figma_status
• figma_eval
• figma_metadata
• figma_tokens
• figma_find
• figma_changes
• figma_history
• figma_library
• figma_export
• figma_place_image
• LOCALFIG_HOME
• LOCALFIG_PORT
• LOCALFIG_OUT

Tested with

Claude Desktop, Claude Code, Cursor, Windsurf, Cline, VS Code

Example client config

{ "mcpServers": { "localfig": { "command": "npx", "args": ["-y", "localfig"] } } }

Notes

Turns the Figma desktop app into a local MCP server, exposing the full Figma Plugin API to AI agents. It reads and writes the open file, including tokens, components, exports, and undo, without a cloud API, token, or quota. One command registers it with Claude Code, Cursor, VS Code, Windsurf, Cline, Gemini CLI, Codex, and other MCP clients.

0 stars on GitHub. Last updated 2026-09-16. Licensed MIT.

Use cases

• Read and write design tokens in the open Figma file
• Create or modify components programmatically from an AI agent
• Export assets and trigger undo operations via the Plugin API

Pros

• Runs fully local with no cloud API, token, or quota
• Zero dependencies and one-command setup for major AI coding tools
• Exposes the complete Figma Plugin API, not a limited subset

Cons

• Requires the Figma desktop app to be running with the target file open
• Limited to the local machine and the currently open file, no cloud collaboration
• No community traction yet (0 stars), so maturity is unproven

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.