This project provides an Excalidraw server that integrates hand-drawn diagramming capabilities into Claude AI conversations via the Model Context Protocol (MCP) Apps. It enables users to generate and interactively edit diagrams directly within the chat interface, enhancing communication beyond static text.

How It Works

The server leverages the MCP Apps framework, allowing AI models to return interactive HTML content. It specifically embeds Excalidraw, a virtual whiteboard tool, enabling the streaming of hand-drawn diagrams with smooth camera controls and fullscreen editing capabilities directly within the Claude chat environment.

Quick Start & Requirements

• Cloud Deployment: Configure Claude.ai settings with Server URL: https://excalidraw-mcp-app.vercel.app/mcp.
• Local Deployment (Claude Desktop):
  • Install via downloaded .mcpb bundle from releases.
  • Alternatively, build from source: git clone https://github.com/antonpk1/excalidraw-mcp-app.git, cd excalidraw-mcp-app, npm install && npm run build.
  • Configure ~/Library/Application Support/Claude/claude_desktop_config.json with the Node.js command and path to the built dist/index.js. Restart Claude Desktop.
• Prerequisites: Node.js (for local build), mcpb tool (for packaging).

Highlighted Details

• Interactive, hand-drawn diagram creation within Claude AI.
• Smooth viewport camera control for navigation.
• Fullscreen editing mode for enhanced usability.
• Supports prompts like "Draw a cute cat using excalidraw" or complex architecture diagrams.

Maintenance & Community

The project outlines a clear process for releasing new versions, involving code changes, version bumping in manifest.json and package.json, rebuilding, packing the .mcpb bundle, and creating GitHub releases. Links to MCP Apps documentation and Excalidraw are mentioned.

Licensing & Compatibility

The project is released under the MIT License, permitting broad use, modification, and distribution, including for commercial purposes, provided the license terms are met.

Limitations & Caveats

Functionality is dependent on the Claude AI platform's MCP App integration. Local setup requires Node.js and manual configuration within Claude Desktop's settings. No specific performance benchmarks or hardware requirements beyond standard Node.js execution are detailed.