MCP -fastapi-containerapp

This MCP server provides a Streamable-Http interface exposing tools such as sports news retrieval (e.g., getting the latest NFL news) via a FastMCP-based implementation. It supports API key authentication and can be extended with additional tools or data sources. The server is designed to be consumed by MCP clients (Agents) built with Autogen, the OpenAI Agent SDK, Semantic Kernel, or other MCP-compatible clients. To use it, deploy the server (locally or to Azure Container Apps) and point your MCP clients to the server's Streamable-Http endpoints. Clients can query for tools, invoke them, and stream results back in real time as news updates or other data.

You can access example MCP clients included in this repository to see how to interact with the server. The Autogen, OpenAI Agent SDK, and Semantic Kernel-based clients demonstrate how to host an MCP Host/Client, discover available MCP tools, and execute them with appropriate prompts or agent logic. Ensure you have a valid API key if your client tooling relies on external data sources (e.g., sports news APIs) and that API key authentication is configured on the server side if you plan to restrict access.

Prerequisites:

• Python 3.10+ and uv (via uvx) installed
• Access to the project repository and the ability to install the MCP server package
• Optional: Docker if you prefer containerized deployment

Step-by-step installation:

1. Install the uv package manager (if not already installed):
   • Follow instructions at https://docs.astral.sh/uv/
2. Install the MCP server package using uvx (from the repository name):
   • uvx joelborellis-mcp-server-fastapi-containerapp
3. Prepare environment variables:
   • Copy example env file: cp .env.sample .env
   • Edit .env to configure MCP authentication, host, port, and any API keys your tools require
4. Sync/install dependencies (per pyproject.toml):
   • uv sync
5. Run the MCP server:
   • uvx joelborellis-mcp-server-fastapi-containerapp
   • Or if a direct node-based approach is used in your environment, follow the project-specific run command provided in the repository's docs

If you prefer containerized deployment, you can build and run a container image and expose the port used by the server. See the repository’s Azure Container Apps instructions for containerized deployment specifics.

Notes and tips:

• The server implements a Streamable-Http MCP interface with API key authentication. Ensure API keys are securely managed and not leaked in logs.
• The environment file (.env) should include endpoints and credentials for any data sources used by the tools (e.g., sports data APIs).
• If you modify tools or add new ones, ensure the MCP tooling discovery works by restarting the server and reloading the available tool list.
• When debugging, check server logs for authentication errors, tool discovery, and stream events. Network access (ports) must be open between MCP clients and the server.
• For container deployments, verify that container-app networking and environment variables map correctly to the container’s runtime environment.