Mcpo

⚡️ mcpo Documentation (English)

Transform any MCP tool into an OpenAPI-compatible HTTP server – instantly.

mcpo is a simple and easy-to-use proxy that converts MCP server commands into standard RESTful OpenAPI interfaces, making your tools "plug-and-play" compatible with LLM agents and applications that expect an OpenAPI server.

No custom protocols. No glue code. No hassle.

🚀 Quick Start

We recommend using uv to start with lightning speed and zero configuration.

uvx mcpo --port 8000 --api-key "top-secret" -- your_mcp_server_command

Alternatively, if you use Python:

pip install mcpo
mcpo --port 8000 --api-key "top-secret" -- your_mcp_server_command

If you need to use an MCP server that supports SSE (Server-Sent Events), specify the server type and endpoint:

mcpo --server-type sse --host your_sse_server_host:your_sse_server_port

Manage multiple tools using a configuration file:

mcpo --config your_config_file.yaml

After starting, visit http://localhost:8000 to view the automatically generated API documentation.

✨ Features

Why choose mcpo over native MCP?

MCP servers typically communicate using raw stdio, which has the following issues:

• Insecurity: Inherently insecure
• Incompatibility: Incompatible with most tools
• Lack of standard features: Such as documentation, authentication, error handling, etc.

mcpo solves all these problems – without any extra work:

• ✅ Instantly compatible with OpenAPI tools, SDKs, and UIs
• 🛡 Increase security and stability with trusted web standards
• 🧠 Automatically generate interactive documentation without any configuration
• 🔌 Use pure HTTP – no sockets, no glue code, no surprises

What seems like "one more step" actually results in fewer steps and better outcomes.

mcpo makes your AI tools usable, secure, and more interoperable – instantly, with zero hassle.

💻 Usage Examples

1. Configuration file your_config_file.yaml:

servers:
  - name: "tool1"
    host: "localhost:1234"
  - name: "tool2"
    host: "anotherhost:5678"

2. Access the documentation page after startup

🔧 Technical Details

Technical Requirements

• Python: Python 3.8 or higher is required
• Dependencies:
  • uvicorn: Used to start the Web server
  • python-multipart: Handles file uploads

Development and Testing

Clone the repository

git clone https://github.com/open-webui/mcpo.git
cd mcpo

Install dependencies

pip install -r requirements.txt

Start testing

uvicorn main:app --reload

Visit http://localhost:8000 to view the automatically generated documentation.

📄 License

Configuration file example

config.yaml:

servers:
  - name: "example_tool"
    host: "127.0.0.1:5000"

Run the command:

mcpo --config config.yaml

🎯 Goals

Build the interoperability of future AI tools!

✨ Let's build an open and interconnected ecosystem of AI tools together!