Open Multi-Agent Canvas is an open-source chat interface for managing multiple AI agents in dynamic conversations, built with Next.js, LangGraph, and CopilotKit. It targets users needing to coordinate agents for tasks like travel planning and research, offering a unified platform for complex workflows.

How It Works

The system leverages LangGraph for orchestrating multi-agent workflows and CopilotKit for building the chat interface. It supports a Multi-Channel Protocol (MCP) for agent communication, allowing integration with various MCP-compatible servers, including local standard I/O commands and external SSE endpoints. This modular approach enables flexible agent deployment and interaction.

Quick Start & Requirements

• Install: cd frontend && pnpm i
• Prerequisites: pnpm, CopilotKit API key, OpenAI API key, LangSmith API key (optional for MCP Agent).
• Setup: Requires renaming .env files and installing dependencies. Running the MCP Agent backend involves poetry install and poetry run langgraph dev.
• Docs: CopilotKit Docs, LangGraph Platform Docs, Model Context Protocol (MCP) Docs.

Highlighted Details

• Supports multiple agents in a single conversation.
• Integrates with custom MCP servers (Standard IO, SSE).
• Includes a general-purpose MCP Agent.
• Built using Next.js, LangGraph, and CopilotKit.

Maintenance & Community

Developed by CopilotKit. Links to documentation are provided.

Licensing & Compatibility

Distributed under the MIT License. Permissive for commercial use and closed-source linking.

Limitations & Caveats

Requires specific API keys for full functionality. The MCP Agent backend setup involves Poetry and LangGraph development server, which may require additional configuration for production deployment.