figma-developer-docs-mcp
# figma-developer-docs-mcp
An MCP (Model Context Protocol) server that provides AI coding assistants with structured access to the complete Figma developer documentation.
## Coverage
| Section | Pages | Content |
|---------|-------|---------|
| Plugin API | 450 | Node types, properties, methods, guides |
| Widget API | 78 | Components, hooks, types, guides |
| REST API | 51 | Endpoints, authentication, webhooks, SCIM |
| Figma MCP Server | 23 | Setup, tools, troubleshooting |
| Code (Figma Make) | 19 | Code layers, properties, guidelines |
| Code Connect | 18 | React, SwiftUI, Compose, CI/CD |
| Embeds | 8 | Embed kit, security, API |
| **Total** | **649** | |
## Usage
### With Claude Code
```bash
claude mcp add figma-developer-docs -- npx figma-developer-docs-mcp
```
### With Claude Desktop
Add to your `claude_desktop_config.json`:
```json
{
"mcpServers": {
"figma-developer-docs": {
"command": "npx",
"args": ["figma-developer-docs-mcp"]
}
}
}
```
### With any MCP client
```bash
npx figma-developer-docs-mcp
```
The server communicates over stdio using the MCP protocol.
## Tools
### `read-figma-developer-docs-index`
Returns the full documentation index — all sections and page listings. Use this first to discover what's available.
### `read-figma-developer-docs-pages`
Reads specific documentation pages by path. Accepts an array of paths from the index.
```
paths: ["/docs/plugins/api/TextNode", "/docs/plugins/working-with-text"]
```
### `search-figma-developer-docs`
Full-text search across all documentation. Returns matching pages with excerpts.
```
query: "createImage"
section: "plugins" // optional: limit to a section
maxResults: 10 // optional
```
## Development
```bash
# Install dependencies
npm install
# Build
npm run build
# Re-scrape documentation (requires internet)
npm run scrape
# Run locally
node dist/index.js
```
## How it works
The documentation is scraped from [developers.figma.com](https://developers.figma.com) and converted to Markdown files organized by section. The MCP server reads these files on demand and serves them through three tools: index, read, and search.
The scraped docs are included in the npm package so the server works offline with no network requests.
## License
MIT
TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose: one reads the index to discover available documentation, one reads specific pages for detailed content, and one searches across all documentation. There is no overlap in functionality, making it easy for an agent to select the right tool based on the task.
All tool names follow a consistent verb_noun pattern with hyphens: 'read-figma-developer-docs-index', 'read-figma-developer-docs-pages', and 'search-figma-developer-docs'. This predictability enhances usability and reduces confusion.
Three tools are well-scoped for the server's purpose of accessing Figma developer documentation, covering discovery, reading, and searching. While slightly minimal, each tool earns its place, and there is no bloat or obvious missing functionality for this focused domain.
The tool set provides complete coverage for the domain of accessing documentation: it allows agents to discover available docs via the index, read specific pages for detailed information, and search across all content when needed. There are no gaps that would cause agent failures in typical documentation workflows.