mcpo

by ericsantos

API/Gateways & Proxies OpenAPI HTTP Proxy RESTful

mcpo is a proxy that exposes any MCP tool as an OpenAPI-compatible HTTP server. It allows your tools to work with LLM agents and apps expecting OpenAPI servers without custom protocols or glue code.

View on GitHub

Last updated: N/A

What is mcpo?

mcpo is a dead-simple proxy that takes an MCP server command and makes it accessible via standard RESTful OpenAPI. It solves the issues of insecurity, incompatibility, and missing standard features associated with raw stdio MCP servers.

How to use mcpo?

Install mcpo using pip or uv, or run it via Docker. Then, use the mcpo command followed by the MCP server command and any necessary flags (like port and API key). You can also use a config file to serve multiple MCP tools.

Key features of mcpo

Exposes MCP tools as OpenAPI servers
Auto-generates interactive documentation
Adds security, stability, and scalability
Works with OpenAPI tools, SDKs, and UIs

Use cases of mcpo

Integrating MCP tools with LLM agents
Making MCP tools accessible to applications expecting OpenAPI servers
Securing and standardizing access to MCP tools
Providing interactive documentation for MCP tools

FAQ from mcpo

Why use mcpo instead of native MCP?

Native MCP servers usually speak over raw stdio, which is inherently insecure, incompatible with most tools, and missing standard features. mcpo solves these issues by exposing MCP tools via OpenAPI.

How do I install mcpo?

You can install mcpo using pip (pip install mcpo), uv (uvx mcpo), or run it via Docker (docker run).

How do I access the OpenAPI schema?

The OpenAPI schema is available at http://localhost:8000/docs (or http://localhost:8000/<tool>/docs when using a config file).

Can I serve multiple MCP tools with mcpo?

Yes, you can serve multiple MCP tools using a config file in the Claude Desktop format.

What are the requirements for using mcpo?

mcpo requires Python 3.8+ and uv is recommended for performance.