mcpo-docker
An example Docker image for mcpo, a tool that exposes MCP (Model Context Protocol) servers as OpenAPI-compatible HTTP endpoints for OpenWebUI.
claude mcp add --transport stdio alephpiece-mcpo-docker docker run -i mcpo
How to use
mcpo-docker exposes MCP servers via an OpenAPI-compatible HTTP surface using an OpenWebUI integration. The image runs one or more MCP servers and assigns each server its own HTTP endpoint that OpenWebUI can call as a tool. Typical endpoints created by mcpo include memory, sequential-thinking, and time, for example http://localhost:8000/memory, http://localhost:8000/sequential-thinking, and http://localhost:8000/time. To connect OpenWebUI, configure a tool connection pointing to the host and port where the container exposes its API (default 8000 in the examples), and OpenWebUI will list the available MCP endpoints as tools you can invoke in chat.
How to install
Prerequisites:\n- Docker and Docker Compose installed on your machine.\n- Git installed if cloning from GitHub.\n\nInstallation steps:\n1) Clone the repository (or use the provided image).\n git clone https://github.com/alephpiece/mcpo-docker.git\n cd mcpo-docker\n\n2) Copy sample configuration files (optional).\n cp config.example.json config.json\n cp docker-compose.example.yml docker-compose.yml\n\n3) Build or pull the mcpo Docker image. You can build locally with:\n docker build -t mcpo .\n # or pull from a registry if available and adapt docker-compose accordingly.\n\n4) Start the container(s) using docker-compose:\n docker compose up -d\n # Wait for the servers to initialize. The endpoints will be available at http://localhost:8000/\n\n5) (Optional) Run a standalone container executing mcpo directly:\n docker run -d -p 8000:8000 mcpo\n\nNotes:\n- Ensure port 8000 is exposed and not in use by another service.\n- If you modify config.json, restart the container to apply changes.\n- The exact image name and docker-compose file might vary; adjust commands accordingly if you rename images or services.
Additional notes
Tips and common issues:\n- If OpenWebUI cannot reach an endpoint, verify the container is running and listening on the expected port (default 8000).\n- Check logs with docker compose logs to see which MCP servers are active and their endpoints.\n- Endpoints correspond to individual MCP servers configured in config.json; removing a server from the config will remove its OpenAPI surface.\n- When using docker-compose, ensure environment variables in config.json are mounted into the container if required.\n- If you see SSL or CORS-related issues in OpenWebUI, verify the server supports the requested origin and that the OpenWebUI tool can reach the host network.\n- For multiple MCP servers, each server gets a distinct endpoint like /memory, /time, /sequential-thinking, etc.
Related MCP Servers
mcpo
A simple, secure MCP-to-OpenAPI proxy server
guidance-for-scalable-model-inference-and-agentic-ai-on-amazon-eks
Comprehensive, scalable ML inference architecture using Amazon EKS, leveraging Graviton processors for cost-effective CPU-based inference and GPU instances for accelerated inference. Guidance provides a complete end-to-end platform for deploying LLMs with agentic AI capabilities, including RAG and MCP
mcp-cyberbro
Using MCP is fun with Cyberbro!
searxng
An MCP sse implementation of the Model Context Protocol (MCP) server integrated with SearXNG for providing AI agents with powerful, privacy-respecting search capabilities.
openwebui -setup
Universal MCPO/MCP bridge for Open Web UI with AI-powered configuration. Automated setup generation, Docker support, beginner-friendly. Any AI assistant can instantly convert MCP configs to working Open Web UI integrations.
agentsource -ext
Access live company and contact data through Explorium's MCP server