openapi s

This MCP server collection exposes a set of OpenAPI-enabled tool endpoints implemented as reference FastAPI-based servers. Each server (for example filesystem, weather, git, memory, sql, and external RAG) runs as its own HTTP API compliant with OpenAPI standards, allowing LLM agents and automation workflows to discover and invoke tool actions in a consistent manner. To use them, start the individual server(s) and direct your OpenAPI clients or MCP tooling to the corresponding base URL and path. The included MCP bridge tooling can connect these OpenAPI servers to MCP-compatible workflows, enabling tool invocation, parameter passing, and response handling across the ecosystem.

Usage patterns include:

• Start specific servers locally (e.g., filesystem and weather) to expose local file operations and weather data as APIs.
• Use the MCP bridge (mcpo-based or python proxy) to translate between MCP tool invocations and OpenAPI endpoints. This lets you integrate these servers into existing MCP pipelines without rewriting tooling.
• Query and call endpoints via standard OpenAPI clients, or route through MCP adapters to maintain a consistent tool invocation surface for agents.

Prerequisites:

• Python 3.9+ installed on your system
• Optional: Docker for running with docker-compose
• Git to clone the repository

1. Clone the repository

• git clone https://github.com/open-webui/openapi-servers.git
• cd openapi-servers

2. Install dependencies for a specific server Each server lives under servers/<server-name> and typically includes a requirements.txt. Install and run as shown for each server you want to expose.

Example for the Filesystem server:

cd servers/filesystem
pip install -r requirements.txt
uvicorn main:app --host 0.0.0.0 --reload

Example for the Weather server:

cd servers/weather
pip install -r requirements.txt
uvicorn main:app --host 0.0.0.0 --reload

3. Run with Docker (optional) If you prefer Docker, each server can be run using a Docker Compose setup where available. For example:

cd servers/filesystem
docker compose up

4. Verify runs Open a browser or API tool and navigate to http://localhost:8000/docs (or the port you specified) to view the OpenAPI docs and interact with the endpoints.

Tips and common considerations:

• Each server may require environment variables (e.g., API keys for external data sources). If needed, add them to the environment when starting the server, for example: export OPENAI_API_KEY=your_key, then run the uvicorn command.
• Ports: The sample commands expose each server on a unique port (8000, 8001, etc.). If you run multiple instances or in Docker, ensure port mappings do not conflict.
• Security: Expose only necessary endpoints and apply authentication as appropriate for production use. OpenAPI docs can be public for testing, but protect sensitive operations.
• Bridging: Use mcpo or a Python proxy to bridge these OpenAPI servers to MCP workflows. This lets MCP-based agents call OpenAPI endpoints transparently.
• Extensibility: The repository is designed to show reference implementations. You can adapt the openapi-based servers to your own tooling stacks (FastAPI, FastOpenAPI, etc.) while keeping the MCP integration intact.