You can use the following production-style README.md for your FastMCP server and client applications.

FastMCP Tools Server

A production-ready FastMCP server exposing the following tools:

• health - Health check endpoint
• say_hello - Returns a greeting message
• generate_uuid - Generates a random UUID

Project Structure

project/
│
├── app/
│   ├── __init__.py
│   └── server.py
│
├── client/
│   └── mcp_client.py
│
├── tests/
│   └── test_mcp_tools.py
│
├── requirements.txt
├── pyproject.toml
└── README.md

---

Prerequisites

• Python 3.14+
• pip

Verify installation:

python --version

Expected:

Python 3.14.x

---

Create Virtual Environment

Windows

python -m venv .venv
.venv\Scripts\activate

Linux / macOS

python -m venv .venv
source .venv/bin/activate

---

Install Dependencies

pip install -r requirements.txt

or

pip install fastmcp pytest pytest-asyncio

---

Start MCP Server

Navigate to the project root folder.

Run:

python app/server.py

Expected output:

INFO: FastMCP server started
INFO: Listening on http://0.0.0.0:8000/mcp

---

Verify Server

Open a new terminal and run:

curl http://localhost:8000/mcp

or

fastmcp list http://localhost:8000/mcp

Expected tools:

health
say_hello
generate_uuid

---

Run Client Application

Example:

python client/mcp_client.py

Expected output:

Available Tools:
- health
- say_hello
- generate_uuid

Health Response:
{'status': 'healthy'}

Hello Response:
{'message': 'Hello Satheesh'}

UUID Response:
{'uuid': 'xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx'}

---

Example Client Code

import asyncio

from fastmcp import Client


async def main():
    async with Client("http://localhost:8000/mcp") as client:

        health = await client.call_tool("health")
        print(health.data)

        hello = await client.call_tool(
            "say_hello",
            {"name": "Satheesh"}
        )
        print(hello.data)

        uuid_result = await client.call_tool(
            "generate_uuid"
        )
        print(uuid_result.data)


if __name__ == "__main__":
    asyncio.run(main())

---

Run Tests

Execute:

pytest -v

Expected:

tests/test_mcp_tools.py::test_health PASSED
tests/test_mcp_tools.py::test_say_hello PASSED
tests/test_mcp_tools.py::test_generate_uuid PASSED

---

Available Tools

health

Request:

await client.call_tool("health")

Response:

{
  "status": "healthy"
}

say_hello

Request:

await client.call_tool(
    "say_hello",
    {
        "name": "Satheesh"
    }
)

Response:

{
  "message": "Hello Satheesh"
}

generate_uuid

Request:

await client.call_tool(
    "generate_uuid"
)

Response:

{
  "uuid": "550e8400-e29b-41d4-a716-446655440000"
}

---

Troubleshooting

List Tools Works but Tool Returns None

Verify the server returns a dictionary:

@mcp.tool
def say_hello(name: str) -> dict:
    return {
        "message": f"Hello {name}"
    }

Avoid returning primitive values such as strings for production APIs.

Connection Refused

Ensure the MCP server is running:

python app/server.py

Verify port:

netstat -ano | findstr 8000

Verify FastMCP Version

pip show fastmcp

Recommended:

fastmcp >= 3.x

---

Production Recommendations

• Run behind Nginx or API Gateway.
• Enable HTTPS.
• Add authentication and authorization.
• Use structured responses (dict/Pydantic models).
• Add logging and monitoring.
• Deploy with Docker/Kubernetes.
• Add retry and timeout handling in clients.

---

License

MIT License

This README should be sufficient for onboarding developers, running the server/client locally, and validating the MCP tools end-to-end.