Open Multi-Agent Canvas
LocalAbout Open Multi-Agent Canvas
Open Multi-Agent Canvas is an open-source multi-agent chat interface created by CopilotKit, letting users manage several agents within one conversation. It is built with Next.js, LangGraph and CopilotKit and targets travel planning, research and general-purpose tasks through MCP servers.
Key features
- 01Manage multiple agents in one conversation
- 02Built-in MCP Agent with configurable MCP servers
- 03Standard IO and SSE connection options
- 04Connects to public MCP servers such as mcp.composio.dev
Use cases
- 01Deep research through connected MCP servers
- 02Travel planning across coordinated agents
- 03General-purpose task handling in one interface
MCP support
The project acts as an MCP client: users add Standard IO or SSE MCP servers in the interface, and the built-in MCP Agent uses their capabilities.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEOpen Multi-Agent Canvas
Open Multi-Agent Canvas, created by CopilotKit is an open-source multi-agent chat interface that lets you manage multiple agents in one dynamic conversation. It's built with Next.js, LangGraph, and CopilotKit to help with travel planning, research, and general-purpose tasks through MCP servers.
Existing Agents
Check out these awesome agents (they live in separate repositories). You can run them separately or deploy them on LangSmith:
Additionally, this project now includes a built-in MCP (Multi-Channel Protocol) Agent:
- MCP Agent: A general-purpose agent capable of handling various tasks through configurable MCP servers.
Copilot Cloud is required to run this project:
Quick Start 🚀
1. Prerequisites
Make sure you have:
2. API Keys
Running the Frontend
Rename the example.env file in the frontend folder to .env:
NEXT_PUBLIC_CPK_PUBLIC_API_KEY=...
Install dependencies:
cd frontend
pnpm i
Need a CopilotKit API key? Get one here.
Then, fire up the Next.js project:
pnpm run build && pnpm run start
MCP Agent Setup
The MCP Agent allows you to connect to various MCP-compatible servers:
-
Configuring Custom MCP Servers:
- Click the "MCP Servers" button in the top right of the interface
- Add servers via the configuration panel:
- Standard IO: Run commands locally (e.g., Python scripts)
- SSE: Connect to external MCP-compatible servers (via Server-Sent Events)
-
Public MCP Servers:
- You can connect to public MCP servers like mcp.composio.dev and mcp.run
Running the MCP Agent Backend (Optional)
Rename the example.env file in the agent folder to .env:
OPENAI_API_KEY=...
LANGSMITH_API_KEY=...
If you want to use the included MCP Agent with the built-in math server:
cd agent
poetry install
poetry run langgraph dev --host localhost --port 8123 --no-browser
Running a tunnel
Add another terminal and select Remote Endpoint.
Then select Local Development.
Once this is done, copy the command into your terminal and change the port to match the LangGraph server 8123
Documentation
License
Distributed under the MIT License. See LICENSE for more info.
Note: This project has been consolidated into the CopilotKit monorepo. The latest version lives at
examples/showcases/multi-agent-canvas. Please open issues and pull requests in the main CopilotKit repository.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.