Excalidraw MCP Server: AI-Powered Live Visual Diagramming

🙏 Acknowledgments This project is based on and extends the excellent work from yctimlin/mcp_excalidraw. Special thanks to the original contributors for creating the foundation that made this enhanced version possible.

A dual-language MCP server that combines Excalidraw canvas manipulation with AI-driven tool access for live diagram creation.

It uses a Python FastMCP server plus a TypeScript canvas server for separation of protocol handling, canvas state, and browser updates.

Quick Links

• What This System Does
• Architecture Overview
• Key Features
• Usage Guide
• Development

Quality & CI

Crackerjack is used as the standard quality-control and CI/CD gate for the Python side of this repo. Pair it with the existing TypeScript build and test checks for cross-stack changes.

Installation via Bodai Marketplace

This repo ships as a Bodai-style Claude Code plugin. Register the local marketplace, then install the plugin — the manifest's mcpServers entry points at the colocated .mcp.json (HTTP transport, port 3032) and exposes the three slash commands in commands/. The TypeScript canvas server still needs to be running for live updates to flow through to the browser. Use claude plugin marketplace add /Users/les/Projects/bodai-plugins && claude plugin install excalidraw --marketplace bodai-plugins after the marketplace entry lands in bodai-plugins/.claude-plugin/marketplace.json.

🚀 What This System Does

• 🎨 Live Canvas: Real-time Excalidraw canvas accessible via web browser
• 🤖 AI Integration: MCP server allows AI agents (like Claude) to create visual diagrams
• ⚡ Real-time Sync: Elements created via MCP API appear instantly on the canvas
• 🔄 WebSocket Updates: Live synchronization across multiple connected clients
• 🏗️ Production Ready: Clean, minimal UI suitable for end users

🎥 Demo Video

See MCP Excalidraw in Action!

Watch how AI agents create and manipulate diagrams in real-time on the live canvas

🏛️ Architecture Overview

sequenceDiagram
    participant AI as AI Agent<br/>(Claude)
    participant MCP as MCP Server<br/>(Python FastMCP)
    participant Canvas as Canvas Server<br/>(Express.js)
    participant WS as WebSocket Server
    participant UI as React Frontend<br/>(Excalidraw)

    AI->>MCP: MCP Tool Call<br/>(create_element)
    activate MCP
    MCP->>MCP: Validate Request<br/>(Pydantic)
    MCP->>Canvas: HTTP POST<br/>/api/elements
    activate Canvas
    Canvas->>Canvas: Store Element<br/>(in-memory Map)
    Canvas->>WS: Broadcast Update
    WS->>UI: WebSocket Message<br/>(element_created)
    deactivate Canvas
    UI->>UI: Render in Excalidraw
    MCP-->>AI: Tool Response
    deactivate MCP

    Note over MCP,Canvas: Auto-management: Python server<br/>monitors and restarts canvas server
    Note over WS,UI: Real-time: All connected clients<br/>receive updates simultaneously

Hybrid Architecture Benefits:

• Python FastMCP: Handles MCP protocol, tool registration, and auto-manages canvas server
• TypeScript Canvas: Express.js API + WebSocket for real-time canvas synchronization
• Auto-Management: Python server monitors and restarts canvas server as needed
• Type Safety: Comprehensive TypeScript definitions ensure consistency across the stack

🌟 Key Features

Modern TypeScript Architecture

• Full TypeScript Migration: Complete type safety for backend and frontend
• Comprehensive Type Definitions: Excalidraw elements, API responses, WebSocket messages
• Strict Type Checking: Enhanced development experience and compile-time error detection
• Type-Safe React Components: TSX components with proper props typing

Real-time Canvas Integration

• Elements created via MCP appear instantly on the live canvas
• WebSocket-based real-time synchronization
• Multi-client support with live updates

Production-Ready Interface

• Clean, minimal UI with connection status
• Simple "Clear Canvas" functionality
• No development clutter or debug information

Comprehensive MCP API

• Element Creation: rectangles, ellipses, diamonds, arrows, text, lines
• Element Management: update, delete, query with filters
• Batch Operations: create multiple elements in one call
• Advanced Features: grouping, alignment, distribution, locking

Robust Architecture

• TypeScript-based Express.js backend with REST API + WebSocket
• React frontend with official Excalidraw package and TypeScript
• Dual-path element loading for reliability
• Auto-reconnection and error handling

📦 Installation & Setup

✅ Quick Start (Recommended)

1. Clone and Setup

git clone https://github.com/lesleslie/excalidraw-mcp.git
cd excalidraw-mcp

# Install Python dependencies
uv sync

# Install Node.js dependencies from package-lock.json (node_modules not in repo)
npm install
npm run build

2. Start the System

# The Python MCP server auto-starts the canvas server
uv run python excalidraw_mcp/server.py

# Or manually start canvas server (optional)
npm run canvas

3. Access the Canvas

Open your browser and navigate to:

http://localhost:3031

📋 Development Setup

# Development mode (TypeScript watch + Vite dev server)
npm run dev

# Or production mode
npm run production

Development Workflow:

1. Setup → uv sync && npm install && npm run build
2. Development → npm run dev (TypeScript watch + Vite dev server)
3. Testing → pytest (Python) + npm test (TypeScript)
4. Quality Checks → uv run ruff check + npm run type-check
5. Iterate → Make changes, tests auto-reload
6. Commit → All quality checks pass

Development Modes:

🔧 Available Scripts

🎯 Usage Guide

For End Users

1. Open the canvas at http://localhost:3031
2. Check connection status (should show "Connected")
3. AI agents can now create diagrams that appear in real-time
4. Use "Clear Canvas" to remove all elements

For AI Agents (via MCP)

The MCP server provides these tools for creating visual diagrams:

Basic Element Creation

// Create a rectangle
{
  "type": "rectangle",
  "x": 100,
  "y": 100,
  "width": 200,
  "height": 100,
  "backgroundColor": "#e3f2fd",
  "strokeColor": "#1976d2",
  "strokeWidth": 2
}

Create Text Elements

{
  "type": "text",
  "x": 150,
  "y": 125,
  "text": "Process Step",
  "fontSize": 16,
  "strokeColor": "#333333"
}

Create Arrows & Lines

{
  "type": "arrow",
  "x": 300,
  "y": 130,
  "width": 100,
  "height": 0,
  "strokeColor": "#666666",
  "strokeWidth": 2
}

Batch Creation for Complex Diagrams

{
  "elements": [
    {
      "type": "rectangle",
      "x": 100,
      "y": 100,
      "width": 120,
      "height": 60,
      "backgroundColor": "#fff3e0",
      "strokeColor": "#ff9800"
    },
    {
      "type": "text",
      "x": 130,
      "y": 125,
      "text": "Start",
      "fontSize": 16
    }
  ]
}

🔌 Integration with Claude Code

✅ Recommended: uvx Configuration

Add this configuration to your Claude Code .mcp.json:

{
  "mcpServers": {
    "excalidraw": {
      "command": "uvx",
      "args": ["excalidraw-mcp"],
      "env": {
        "EXPRESS_SERVER_URL": "http://localhost:3031"
      }
    }
  }
}

Alternative: Local Development Configuration

For local development, use the console-script entry point installed by uv sync:

{
  "mcpServers": {
    "excalidraw": {
      "command": "uv",
      "args": ["run", "excalidraw-mcp"],
      "cwd": "<absolute-path-to-cloned-excalidraw-mcp>"
    }
  }
}

Important: Replace <absolute-path-to-cloned-excalidraw-mcp> with the actual absolute path to your cloned repository. Running excalidraw_mcp/server.py directly is not a supported entry point — use the excalidraw-mcp console script (uv run excalidraw-mcp) or uv run python -m excalidraw_mcp instead.

🔧 Integration with Other Tools

Cursor IDE

Add to your .cursor/mcp.json:

{
  "mcpServers": {
    "excalidraw": {
      "command": "uvx",
      "args": ["excalidraw-mcp"],
      "env": {
        "EXPRESS_SERVER_URL": "http://localhost:3031"
      }
    }
  }
}

VS Code MCP Extension

For VS Code MCP extension, add to your settings:

{
  "mcp": {
    "servers": {
      "excalidraw": {
        "command": "uvx",
        "args": ["excalidraw-mcp"],
        "env": {
          "EXPRESS_SERVER_URL": "http://localhost:3031"
        }
      }
    }
  }
}

🛠️ Environment Variables

Note: PORT, HOST, and DEBUG are not consumed by the Python MCP server. The canvas server (TypeScript, src/server.ts) reads PORT (default 3031) for its Express listener; the Python MCP HTTP transport listens on a fixed 3032 (see excalidraw_mcp/__main__.py). Use EXPRESS_SERVER_URL to point the Python server at a non-default canvas-server host/port, and LOG_LEVEL to control log verbosity.

📊 API Endpoints

The canvas server provides these REST endpoints:

🎨 MCP Tools Available

Element Management

• create_element - Create any type of Excalidraw element
• update_element - Modify existing elements
• delete_element - Remove elements
• query_elements - Search elements with filters

Batch Operations

• batch_create_elements - Create complex diagrams in one call

Element Organization

• group_elements - Group multiple elements
• ungroup_elements - Ungroup element groups
• align_elements - Align elements (left, center, right, top, middle, bottom)
• distribute_elements - Distribute elements evenly
• lock_elements / unlock_elements - Lock/unlock elements

Resource Access

• get_resource - Access scene, library, theme, or elements data

WebSocket Monitoring

The MCP server also exposes five WebSocket-administration tools (registered by excalidraw_mcp/mcp/websocket_tools.py). They are available whenever the WebSocket collaboration server is enabled:

• websocket_health_check - Liveness probe for the WebSocket server
• websocket_get_status - Active connections, rooms, and per-room subscriber counts
• websocket_list_rooms - Enumerate active diagram/cursor/presence/global rooms
• websocket_broadcast_test_event - Publish a synthetic event to a room (development use)
• websocket_get_metrics - Throughput, rate-limit counters, and connection metrics

🏗️ Development Architecture

Frontend (frontend/src/)

• React + TypeScript: Modern TSX components with full type safety
• Vite Build System: Fast development and optimized production builds
• Official Excalidraw: @excalidraw/excalidraw package with TypeScript types
• WebSocket Client: Type-safe real-time element synchronization
• Clean UI: Production-ready interface with proper TypeScript typing

Canvas Server (src/server.ts → dist/server.js)

• TypeScript + Express.js: Fully typed REST API + static file serving
• WebSocket: Type-safe real-time client communication
• Element Storage: In-memory with comprehensive type definitions
• CORS: Cross-origin support with proper typing
• Middleware: Custom Express middleware components
• Storage: Element storage implementations
• WebSocket: Dedicated WebSocket server components
• Utils: Utility functions for various operations

MCP Server (excalidraw_mcp/server.py)

• Python FastMCP: Python-based Model Context Protocol implementation
• Canvas Sync: HTTP requests to canvas server for element synchronization
• Element Management: Full CRUD operations for Excalidraw elements
• Batch Support: Complex diagram creation through batch operations
• Process Management: Canvas server lifecycle management
• HTTP Client: Async HTTP client with retry mechanisms
• Configuration: Centralized configuration management

Type System (src/types.ts)

• Excalidraw Element Types: Complete type definitions for all element types
• API Response Types: Strongly typed REST API interfaces
• WebSocket Message Types: Type-safe real-time communication
• Server Element Types: Enhanced element types with metadata

🐛 Troubleshooting

flowchart TD
    Start[Issue Detected] --> IssueType{What's the problem?}

    IssueType -->|Canvas Not Loading| CanvasCheck{Build succeeded?}
    IssueType -->|Elements Not Syncing| SyncCheck{Canvas server<br/>running?}
    IssueType -->|WebSocket Issues| WSVerify{Check console<br/>for errors}
    IssueType -->|Build Errors| BuildClean{Node version<br/>OK?}
    IssueType -->|Python Errors| DepCheck{uv sync<br/>succeeded?}

    CanvasCheck -->|No| A1[Run npm run build]
    CanvasCheck -->|Yes| A2{Server running?}

    A2 -->|No| A3[Check port 3031<br/>lsof -i :3031]
    A2 -->|Yes| A4{Health check OK?}

    A4 -->|No| A5[Review console logs]
    A4 -->|Yes| A6[Check firewall]

    SyncCheck -->|No| B1[Python auto-starts<br/>canvas server]
    SyncCheck -->|Yes| B2{EXPRESS_SERVER_URL<br/>set?}

    B2 -->|No| B3[Set env variable]
    B2 -->|Yes| B4{/health OK?}

    B4 -->|No| B5[Check canvas<br/>server logs]
    B4 -->|Yes| B6[Verify WebSocket<br/>connection]

    WSVerify -->|Errors found| C1[Check firewall<br/>settings]
    WSVerify -->|No errors| C2[Refresh browser<br/>page]

    BuildClean -->|No| D1[Delete node_modules<br/>npm install]
    BuildClean -->|Yes| D2[Run type-check<br/>npm run type-check]

    DepCheck -->|No| E1[Run uv sync]
    DepCheck -->|Yes| E2{Python 3.13+?}

    E2 -->|No| E3[Install Python 3.13]
    E2 -->|Yes| E4[Check uv<br/>installation]

    style Start fill:#e3f2fd
    style A1 fill:#c8e6c9
    style A6 fill:#ffcdd2
    style B1 fill:#c8e6c9
    style E4 fill:#fff9c4

Canvas Not Loading

• Ensure npm run build completed successfully
• Verify canvas server is running on port 3031
• Python MCP server auto-starts canvas server - check console for errors

Elements Not Syncing

• Python server automatically manages canvas server
• Verify EXPRESS_SERVER_URL points at a reachable canvas server
• Verify canvas server health at http://localhost:3031/health

WebSocket Connection Issues

• Check browser console for WebSocket errors
• Ensure no firewall blocking WebSocket connections
• Try refreshing the browser page

Build Errors

• Delete node_modules and run npm install
• Check Node.js version (requires 16+)
• Run npm run type-check to identify TypeScript issues
• Run uv sync to update Python dependencies

Python Dependencies

• Use uv sync to install/update Python dependencies
• Ensure Python 3.13+ is installed
• Check uv --version to verify uv installation

📋 Project Structure

excalidraw-mcp/
├── examples/                 # Usage examples
├── excalidraw_mcp/           # Python FastMCP server
│   ├── monitoring/           # Monitoring and health check utilities
│   ├── server.py             # Main MCP server (Python)
│   ├── config.py             # Configuration management
│   ├── element_factory.py    # Element creation utilities
│   ├── http_client.py        # HTTP client for canvas server
│   ├── process_manager.py    # Canvas server lifecycle management
│   ├── retry_utils.py        # Retry mechanisms for failed operations
│   ├── cli.py                # Command-line interface
│   ├── mcp_tools.py          # MCP tool implementations
│   ├── __init__.py           # Package initialization
│   └── __main__.py           # Main entry point
├── frontend/                 # React frontend
│   ├── src/
│   │   ├── App.tsx           # Main React component (TypeScript)
│   │   └── main.tsx          # React entry point (TypeScript)
│   └── index.html            # HTML template
├── src/                      # TypeScript canvas server
│   ├── middleware/           # Express middleware components
│   ├── storage/              # Element storage implementations
│   ├── utils/                # Utility functions
│   ├── websocket/            # WebSocket server components
│   ├── config.ts             # Server configuration
│   ├── server.ts             # Express server + WebSocket (TypeScript)
│   └── types.ts              # Type definitions
├── dist/                     # Compiled TypeScript output
│   ├── server.js             # Compiled canvas server
│   ├── server.d.ts           # Server type definitions
│   ├── server.js.map         # Server source maps
│   ├── types.js              # Compiled type definitions
│   ├── types.d.ts            # Type definition files
│   ├── types.js.map          # Type definition source maps
│   ├── utils/                # Compiled utilities
│   ├── assets/               # Frontend assets
│   └── frontend/             # Built React frontend
├── tests/                    # Python test suite
│   ├── unit/                 # Unit tests
│   ├── integration/         # Integration tests
│   ├── security/             # Security tests
│   ├── performance/         # Performance tests
│   └── e2e/                  # End-to-end tests
├── .github/                  # GitHub configurations
├── pyproject.toml            # Python project configuration
├── package.json             # Node.js dependencies and scripts
├── tsconfig.json             # TypeScript configuration
└── README.md                 # This file

📦 Package Naming

This project uses consistent naming across different contexts:

• Python Package: excalidraw_mcp (underscore) - used in imports and Python module references
• PyPI Distribution: excalidraw-mcp (hyphen) - used for uvx excalidraw-mcp and pip installation
• npm Package: excalidraw-mcp (hyphen) - used for Node.js dependencies
• MCP Server Name: excalidraw - used in .mcp.json configuration

Example Usage:

# Python imports (underscore)
from excalidraw_mcp.server import main

# Shell commands (hyphen)
uvx excalidraw-mcp
pip install excalidraw-mcp

🧪 Testing & Quality Assurance

Coverage Requirements

• Python: 85% minimum test coverage (enforced by pytest)
• TypeScript: 70% minimum test coverage (enforced by Jest)

Running Tests

# Python tests with coverage
pytest --cov=excalidraw_mcp --cov-report=html
pytest --cov=excalidraw_mcp --cov-report=term-missing

# TypeScript tests with coverage
npm run test:coverage

# Run all tests
pytest && npm test

# Specific test categories
pytest tests/unit/                  # Python unit tests
pytest tests/integration/           # Python integration tests
pytest -m security                 # Security tests
pytest -m performance              # Performance benchmarks

npm run test:unit                   # TypeScript unit tests
npm run test:integration            # TypeScript integration tests

Quality Standards

This project enforces strict quality standards:

• All code must pass type checking (Pyright for Python, TSC for TypeScript)
• Security scanning with Bandit for Python
• Linting and formatting with Ruff (Python) and built-in TypeScript rules
• Comprehensive test coverage as specified above

📦 Publishing & Distribution

PyPI Distribution

The Python package is published to PyPI as excalidraw-mcp:

# Install from PyPI (latest published release)
pip install excalidraw-mcp

# Use with uvx (recommended)
uvx excalidraw-mcp

The pyproject.toml/CHANGELOG.md version may briefly lead the latest PyPI release while a publish is in flight; in that window prefer pip install -e . from a fresh clone over pip install excalidraw-mcp.

Local Development

For local development and testing:

# Install in editable mode
pip install -e .

# Or use UV for development
uv sync
uv run python excalidraw_mcp/server.py

Version Management

• Semantic versioning (SemVer) is used
• Version is managed in both pyproject.toml and package.json
• Releases are tagged in git with version numbers

🔮 Development Roadmap

• ✅ Python FastMCP Architecture: Modern hybrid implementation with auto-management
• ✅ TypeScript Canvas Server: Complete type safety for enhanced development experience
• ✅ Comprehensive Testing: Security, performance, and integration test suites
• 🔧 Enhanced Features: Additional MCP tools and canvas capabilities
• 🔧 Performance Optimization: Real-time sync improvements
• 🔧 Docker Deployment: Containerized deployment options

🤝 Contributing

We welcome contributions! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

📝 License

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

🙏 Acknowledgments

• Excalidraw Team - For the amazing drawing library
• MCP Community - For the Model Context Protocol specification