figma-opencode-mcp
Local-first MCP server that connects AI coding agents to the currently open Figma file through a local plugin bridge, requiring no Figma API token.
README
figma-opencode-mcp
Local-first MCP server for OpenCode/Codex that connects AI coding agents to the currently open Figma file through a local plugin bridge.
No Figma API token required. No Figma REST API in default mode. All communication stays on your machine.
Version: v0.2.2 ā fix docs URL in CLI help. Run with a single npx command.
Key Features
- š No Figma API token required in default mode
- š No data leaves your machine ā local WebSocket bridge
- š Read current selection ā get node data from Figma
- š Export selection as JSON ā structured data for design-to-code
- āļø Create frames and text ā basic write operations
- š Audit selection ā design system heuristics (spacing, typography, layout)
- š§© Works with OpenCode and Codex via MCP stdio transport
- š¦ Run with a single
npxcommand ā no clone or install needed
Important Limitations
ā ļø This does not magically access any private Figma file by URL.
- You still need to open the Figma file in Figma.
- You must run the local plugin bridge inside the file.
- The plugin can only access what Figma Plugin API allows in the current file/session.
Architecture
OpenCode/Codex ā MCP stdio ā WebSocket (127.0.0.1:3845) ā Plugin UI ā Plugin Main ā Figma API
See docs/architecture.md for details.
Quick Start
Option A: Install with npx (recommended)
Run the MCP server directly ā no clone or local build needed:
npx -y figma-opencode-mcp@latest
Configure OpenCode/Codex
OpenCode:
opencode mcp add figma-opencode-mcp -- npx -y figma-opencode-mcp@latest
Codex:
codex mcp add figma-opencode-mcp -- npx -y figma-opencode-mcp@latest
.mcp.json (any MCP client):
{
"mcpServers": {
"figma-opencode-mcp": {
"command": "npx",
"args": ["-y", "figma-opencode-mcp@latest"]
}
}
}
Note: The npm/npx command only installs the MCP server. You still need to import and run the Figma plugin separately (see step 4).
Option B: Local clone (for development)
git clone <repo-url> figma-opencode-mcp
cd figma-opencode-mcp
npm install
npm run build
Start MCP Server:
# Development mode (hot reload)
npm run dev
# Production mode
node dist/index.js
Local MCP config:
{
"mcpServers": {
"figma-opencode-mcp": {
"command": "node",
"args": ["/ABSOLUTE/PATH/TO/figma-opencode-mcp/dist/index.js"]
}
}
}
4. Run Figma Plugin (required for both options)
- Build the plugin:
npm run build:plugin(skip if using npx ā download the plugin from releases) - In Figma: Plugins ā Development ā Import plugin from manifest
- Select
plugin/manifest.json - Run the plugin in your Figma file
- The plugin will connect to
ws://127.0.0.1:3845 - Back in OpenCode/Codex, call
figma_ping
Tools
| Tool | Description |
|---|---|
figma_ping |
Check connection status |
figma_get_selection |
Get selected nodes |
figma_get_current_page |
Get current page info |
figma_get_node_tree |
Get node tree |
figma_export_selection_json |
Export as structured JSON |
figma_create_frame |
Create a frame |
figma_create_text |
Create a text node |
figma_audit_selection |
Audit for design issues |
See docs/tools.md for detailed documentation.
Example Prompts
Use figma_ping to check connection.
Use figma_get_selection to read my current selected frame.
Use figma_export_selection_json and generate a React/Vite component.
Use figma_audit_selection and suggest design-system improvements.
Project Structure
figma-opencode-mcp/
āāā src/ # MCP server source
ā āāā index.ts # Entry point
ā āāā mcp/ # MCP layer (server, tools, schemas, handlers)
ā āāā bridge/ # WebSocket bridge (WS server, client, pending requests)
ā āāā shared/ # Shared types (protocol, errors, logger, serialize)
āāā plugin/ # Figma plugin (manifest, code, UI, types)
āāā docs/ # Documentation
āāā examples/ # MCP config examples
āāā scripts/ # Dev/build/verify scripts
āāā README.md
Security
- No token, no REST API in default mode.
- WebSocket binds only to
127.0.0.1. - No data sent to external servers.
- No eval or shell execution.
- No
.envwith secrets.
See docs/security.md.
Roadmap
v0.2.0 ā
- npm package packaging with
npxsupport --versionand--helpCLI flags- Plugin packaging (
plugin.zip) - Documentation updates for npx install
v0.3.0
figma_extract_design_tokensfigma_generate_react_specfigma_compare_code_with_designfigma_create_componentfigma_apply_auto_layoutfigma_export_svg_png
v0.4.0
- Integration profile for
opencode-power-kit /figma-setupslash command/figma-to-reactworkflow/figma-auditworkflowfigma-design-systemskill
v1.0.0
- Stable no-token local bridge
- Better plugin session management
- Better schema docs
License
MIT
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.