n00bvn/CanvasMCPClient
View on GitHub ↗Canvas MCP Client is an open-source, self-hostable dashboard application built around an infinite, zoomable, and pannable canvas. It provides a unified interface for interacting with multiple MCP (Model Context Protocol) servers through a flexible, widget-based system.
77 ★15 forksTypeScriptUpdated 6mo ago
What you need to know
Self-hostable infinite-canvas dashboard application with integrated MCP client support for managing multiple MCP servers through widgets.
Install
git clone https://github.com/n00bvn/CanvasMCPClient.git && cd CanvasMCPClient docker-compose up -d Manual: backend (python -m venv, pip install, uvicorn main:app --port 8081) plus frontend (npm install, npm run dev)
Usage
- •Access the app at http://localhost:3031 (backend API at :8081)
- •Add MCP server configurations in the MCP Servers sidebar section and test connections
- •Set up AI providers in AI Config, then create dashboards with widgets
Key features
- ✓Infinite, zoomable and pannable canvas with grid snap and persistent viewport state
- ✓12+ widgets including sticky notes, to-do lists, AI chat, kanban, spreadsheet, markdown editor, video player and weather
- ✓MCP server management supporting stdio, HTTP and SSE transports with real-time connection status
- ✓Multi AI provider support (OpenAI, Anthropic, Ollama, Google), template system and local SQLite storage
- ✓No-code widget builder and privacy-first local data architecture
Best for
AI power users wanting a visual, widget-based dashboard to interact with multiple MCP servers.
Caveats
- ⚠This project is an MCP client (dashboard), not an MCP server
- ⚠Roadmap items such as oAuth remote MCP servers and real-time collaboration are not yet implemented
Platforms: macOS · Linux · Windows
Documentation ↗Reviewed 2026-08-11
Topics
- Stars
- 77★
- Forks
- 15
- Language
- TypeScript
- License
- MIT
- Created
- 2025-10-15
- Last push
- 2026-02-11