dockerized -template

This MCP server is a Python-based implementation of the Model Context Protocol (MCP) packaged to run inside a Docker container. It uses Streamable HTTP for real-time, stateless transportation of MCP interactions, allowing LLMs to query data and invoke tools exposed by the server in a scalable manner. The repository template demonstrates how to build, run, and validate an MCP server that exposes a simple tool via the MCP framework. You can access the server at the configured endpoint (by default http://localhost:8080/mcp) once the container is up. The included example shows a Python tool defined with @mcp.tool() that adds two numbers; clients can invoke this tool through MCP requests and receive streamed responses as data becomes available. This setup is well-suited for deployment with serverless or containerized environments where stateless transports are preferred for efficiency and scalability.

Prerequisites:

• Docker and Docker Compose installed on your system
• Basic familiarity with Docker workflows

Installation steps:

1. Clone the repository or pull the template docker image you plan to use.

2. If using the Docker-based template, ensure you have a compatible Docker image name. Update the image reference if needed in your environment (the example uses zantis-dockerized-mcp-server-template).

3. Start the server with Docker Compose or a direct Docker run as appropriate for your environment. If using Docker Compose, you can use:

docker-compose up --build

4. If you want to run the server directly with Python (outside Docker) follow these steps:

# Setup Python environment and dependencies
python3 -m venv venv
source venv/bin/activate
pip install -r src/requirements.txt

# Run the server
python src/server.py

5. Verify the server is reachable at http://localhost:8080/mcp or the port you configured. You can also inspect logs to confirm Streamable HTTP is active and the MCP tool is registered.

Notes and tips:

• This template relies on Streamable HTTP for transport; ensure your client tooling supports streaming responses from MCP.
• If you customize tools, decorate functions with @mcp.tool() and provide clear type hints for arguments and return values to facilitate automatic client generation.
• When deploying with Docker, consider exposing the container port 8080 to your host or orchestrator and setting appropriate environment variables for port and endpoint paths.
• Common issues: mismatched API versions between the Python MCP SDK and your client, incorrect tool registration, or firewall rules blocking port 8080. Check MCP and Python SDK docs for version compatibility.
• You can extend the docker template by editing the Dockerfile and server.py to add more tools or data sources as needed.