MCP Server with FastAPI (Python)

A hands-on implementation of a Model Context Protocol (MCP) server built with Python using the official MCP SDK and FastAPI/Starlette.

What is MCP?

Model Context Protocol (MCP) is an open standard that enables AI assistants (like Claude, Kiro, Cursor) to connect with external tools, data sources, and APIs through a unified interface — similar to how USB-C standardized device connectivity.

┌─────────────┐         ┌─────────────┐         ┌──────────────┐
│  AI Client  │  ←───→  │  MCP Server │  ←───→  │  Your Tools/ │
│(Claude/Kiro)│  JSON   │  (This Repo)│         │  Data/APIs   │
└─────────────┘         └─────────────┘         └──────────────┘

Related MCP server: MCP Demo Server

Features

• Tools: Functions that AI clients can discover and call

  • greet — Returns a greeting for a given name

  • add_numbers — Adds two numbers and returns the result

  • get_current_time — Returns the current date and time

• SSE Transport: Server-Sent Events based communication (HTTP)

• Test Client: A Python client that connects to the server and calls all tools

Tech Stack

• Python 3.13+

• MCP Python SDK (v1.9)

• FastAPI / Starlette

• Uvicorn (ASGI server)

• SSE (Server-Sent Events) transport

Getting Started

Prerequisites

• Python 3.10+

• pip

Installation

git clone https://github.com/MuhammadIshaqSkd/mcp-fastapi-tutorial.git
cd mcp-fastapi-tutorial

# Create virtual environment
python -m venv venv

# Activate (Windows)
.\venv\Scripts\activate

# Activate (macOS/Linux)
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

Running the Server

python server.py

Server starts on http://127.0.0.1:8080 with SSE endpoint at /sse.

Testing with the Client

Open a second terminal (keep the server running):

python test_client.py

Expected output:

==================================================
MCP Client - Connecting to server...
==================================================

✅ Connection successful!

📋 Available Tools:
------------------------------
  🔧 greet: Greet someone by name and return a greeting message.
  🔧 add_numbers: Return the sum of two numbers.
  🔧 get_current_time: Return the current date and time.

🎯 Calling 'greet' tool...
  Result: Assalam-o-Alaikum, Mishaq! MCP Server se aapko salam!

🎯 Calling 'add_numbers' tool...
  Result: 15 + 27 = 42

🎯 Calling 'get_current_time' tool...
  Result: Current time: 2026-06-22 18:15:09

==================================================
🎉 All tools called successfully!
==================================================

Project Structure

mcp-fastapi-tutorial/
├── server.py           # MCP Server — registers and exposes tools via SSE
├── test_client.py      # MCP Client — connects to server, lists & calls tools
├── requirements.txt    # Python dependencies
└── README.md

How It Works

1. Server registers tools using @mcp.tool() decorator

2. Client connects via SSE to /sse endpoint

3. Client calls list_tools() to discover available tools

4. Client calls call_tool("tool_name", {args}) to execute a tool

5. Server executes the function and returns the result

Key Concepts Demonstrated

Roadmap

• Basic MCP server with tools

• Add Resources (context/data providers for AI)

• Add Prompts (reusable prompt templates)

• Streamable HTTP transport

• Authentication & authorization

• Deploy to cloud

Resources

• MCP Specification

• MCP Python SDK

• FastMCP Documentation

License

MIT