Vextra MCP Server

A server based on Model Context Protocol (MCP) for processing and parsing Vextra/Figma/Sketch/SVG design files.

Overview

Vextra MCP Server is a powerful design file processing tool that provides AI assistants (like Cursor) with the ability to access and manipulate design files through the MCP protocol. It supports multiple design file formats including Vextra, Figma, Sketch, and SVG.

Related MCP server: Figma MCP Server

Key Features

• 🎨 Multi-format Support: Supports .vext, .sketch, .fig, .svg file formats

• 📊 Layout Information Extraction: Retrieves detailed layout and structure information from design files

• 🖼️ Image Rendering: Renders design nodes as images

• 📋 Page Information Query: Retrieves page structure and metadata

• 🔄 Real-time Communication: Supports SSE (Server-Sent Events) and HTTP communication

• 🌐 Cursor Integration: Seamlessly integrates with Cursor AI assistant

Quick Start

Install Dependencies

npm install

Development Mode

npm run dev

Build Production Version

npm run build
npm start

Configuration

1. Configure Cursor

Add the following MCP configuration in Cursor:

{
  "Vextra MCP": {
    "url": "http://localhost:8080/sse",
    "messageUrl": "http://localhost:8080/messages",
    "type": "sse",
    "headers": {
      "Accept": "text/event-stream",
      "Cache-Control": "no-cache"
    }
  }
}

2. Server Configuration

The server starts by default at http://localhost:8080 and supports the following endpoints:

• GET /sse - SSE connection endpoint

• POST /messages - HTTP message endpoint

API Tools

1. Get Vextra Data

Retrieves layout information and structure data from design files.

Parameters:

• filePath (required): Local file path

• pageId (optional): Page ID

• nodeId (optional): Node ID

• depth (optional): Traversal depth limit

Examples:

Get complete file information
/path/to/file.sketch

Get specific page
/path/to/file.sketch/pageId

Get specific node
/path/to/file.sketch/pageId/nodeId

2. Get Vextra Images

Renders design nodes as images.

Parameters:

• filePath (required): Local file path

• pageId (required): Page ID

• nodeIds (required): Array of node IDs

• format (optional): Image format (png, jpg, svg)

• scale (optional): Scale ratio

3. Get Vextra Pages Info

Retrieves page structure and metadata from files.

Parameters:

• filePath (required): Local file path

Returns:

• Page IDs and names

• Node count statistics

• Page structure overview

Usage Examples

Basic Usage

In Cursor, you can use it like this:

Parse design file
/path/to/file.sketch

Get specific page information
/path/to/file.sketch/pageId

Extract all images from a page
Extract all images from the first page of /path/to/file.sketch

Advanced Usage

Limit depth when analyzing large files
Get layout information from /path/to/large-file.sketch with depth limit of 2 layers

Batch processing
Get thumbnails for all pages in /path/to/design.fig

Project Structure

src/
├── data/           # Data processing modules
│   ├── export/     # Data export related
│   ├── source/     # Data source processing
│   └── vextra.ts   # Vextra service core
├── mcp/            # MCP tool implementations
│   ├── get_vextra_data.ts
│   ├── get_vextra_images.ts
│   └── get_vextra_pagesinfo.ts
├── middlewares/    # Middlewares
└── server.ts       # Server entry point

Tech Stack

• Node.js + TypeScript - Core runtime environment

• Express.js - Web server framework

• @modelcontextprotocol/sdk - MCP protocol support

• @kcaitech/vextra-core - Vextra core library

• skia-canvas - Image rendering engine

• Rollup - Build tool

Development Guide

Environment Requirements

• Node.js >= 16.0.0

• npm >= 8.0.0

Development Scripts

npm run dev       # Development mode
npm run build     # Build production version
npm start         # Start production server

License

This project is licensed under the AGPL-3.0 License - see the LICENSE file for details.

Author

KCai Technology

Related Links

• Vextra Official Website

• Model Context Protocol

• Cursor AI

If you have any questions or suggestions, please contact us through GitHub Issues.