mcpo+OpenWebUI

mcpo+OpenWebUI is an MCP server proxy that converts Model Context Protocol tools into OpenAPI-compatible HTTP servers. It connects standard MCP servers, Server-Sent Events (SSE) servers, and streamable HTTP servers directly to web platforms like OpenWebUI and other REST-compliant AI clients. Software developers, system architects, and AI engineers use this proxy to bridge native protocol tools with web-based user interfaces and API gateways. The server automatically exposes interactive OpenAPI documentation, allowing users to inspect and test endpoints directly through standard browser interfaces. By functioning as an intermediary layer, it handles protocol translation while enforcing API key security and custom port configurations. It enables environments that lack native MCP client support to trigger tool functions via standard HTTP requests, streamlining interoperability between modern AI agent tooling and broader web ecosystems.

Category: AI & LLM Tooling

Tags: http, integration, interoperability, openapi, proxy

Visit mcpo+OpenWebUI

How to install and configure mcpo+OpenWebUI

1. Ensure Python 3.8+ or Docker is installed on your system. 2. Install mcpo via pip using pip install mcpo, run via uv with uvx mcpo, or pull the Docker image using docker run -p 8000:8000 ghcr.io/open-webui/mcpo:main. 3. Configure your downstream MCP server command using the --config flag or inline command syntax. For a configuration file, define your server under mcpServers specifying command, args, or type (such as sse or streamable_http). 4. Start the proxy by specifying the port and API key: mcpo --port 8000 --api-key "your-key" -- your_mcp_server_command. 5. Access the generated OpenAPI documentation at http://localhost:8000/docs to verify endpoints and connect OpenWebUI or your preferred HTTP client.

What you can do with mcpo+OpenWebUI

• Exposing local MCP tools to web interfaces like OpenWebUI through OpenAPI-compliant HTTP endpoints. * Bridging SSE and streamable HTTP servers to standard REST clients requiring predictable request-response cycles. * Generating interactive OpenAPI documentation automatically for inspecting, testing, and debugging MCP server capabilities. * Securing downstream Model Context Protocol tool execution behind an authenticated HTTP proxy with API keys.

Key facts

• Open Source
• https://github.com/open-webui/mcpo
• AI & LLM Tooling, Developer Tools & Code Intelligence
• http, integration, interoperability, openapi, proxy

Part of MCP Servers

How do I install mcpo+OpenWebUI?

You can install mcpo using Python pip with the command pip install mcpo. Alternatively, you can run it directly without prior installation using uvx mcpo, or deploy it as a container using the official ghcr.io/open-webui/mcpo Docker image.

What can mcpo+OpenWebUI do?

mcpo converts MCP tools into OpenAPI-compatible HTTP servers. It supports standard local MCP servers, SSE servers, and streamable HTTP servers, while providing automatic Swagger API documentation at the docs endpoint and securing endpoints with API keys.

What server types does mcpo support?

mcpo supports three main server configurations: standard MCP servers executed via local command-line processes, remote Server-Sent Events compatible servers, and streamable HTTP servers configured with direct URLs.

Is mcpo+OpenWebUI open source?

Yes, mcpo is an open-source project hosted on GitHub by OpenWebUI. Developers can inspect the source code, contribute pull requests, and deploy it freely in their own environments.