FastMCP.SimpleDemo

Perfect — let’s walk through building a FastMCP server with a simple temperature conversion tool, exposing it over HTTP transport, and then connecting it in AI Toolkit for VS Code.

---

🛠 Step 1: Project Setup

1. Create a new folder, e.g. FastMCP.SimpleDemo.
2. Initialize a Python environment:

python -m venv .venv
source .venv/bin/activate   # or .venv\Scripts\activate on Windows
pip install fastmcp

3. Check fastmcp version

fastmcp --version

output:

FastMCP version:                                                                     2.14.0
MCP version:                                                                         1.24.0
Python version:                                                                      3.11.9
Platform:                                                         Windows-10-10.0.26100-SP0
FastMCP root path: C:\Users\ch15p\AppData\Local\Programs\Python\Python311\Lib\site-packages

---

🌡 Step 2: Define the Temperature Conversion Tool

FastMCP lets you define tools that can be exposed to AI agents.
Here’s a simple tool for Celsius ↔ Fahrenheit conversion:

# my_server.py

from fastmcp import FastMCP

mcp = FastMCP("Temp Conversion MCP Server")

@mcp.tool
def celsius_to_fahrenheit(celsius: float) -> float:
    """Convert Celsius to Fahrenheit."""
    return (celsius * 9/5) + 32

@mcp.tool
def fahrenheit_to_celsius(fahrenheit: float) -> float:
    """Convert Fahrenheit to Celsius."""
    return (fahrenheit - 32) * 5/9

if __name__ == "__main__":
    # Expose via HTTP transport
    #mcp.run(host="0.0.0.0", port=8001)
    mcp.run(transport="http", port=8001)

---

🌐 Step 3: Run the Server

Start your server:

fastmcp run tempconversion-mcp-http.py:mcp --transport http --port 8001

output:

╭──────────────────────────────────────────────────────────────────────────────╮
│                                                                              │
│                         ▄▀▀ ▄▀█ █▀▀ ▀█▀ █▀▄▀█ █▀▀ █▀█                        │
│                         █▀  █▀█ ▄▄█  █  █ ▀ █ █▄▄ █▀▀                        │
│                                                                              │
│                                FastMCP 2.14.0                                │
│                                                                              │
│                                                                              │
│                  🖥  Server name: Temp Conversion MCP Server                 │
│                                                                              │
│                  📦 Transport:   HTTP                                        │
│                  🔗 Server URL:  http://127.0.0.1:8001/mcp                   │
│                                                                              │
│                  📚 Docs:        https://gofastmcp.com                       │
│                  🚀 Hosting:     https://fastmcp.cloud                       │
│                                                                              │
╰──────────────────────────────────────────────────────────────────────────────╯
[12/14/25 11:19:02] INFO     Starting MCP server 'Temp Conversion MCP Server' with transport 'http' on                server.py:2582
                             http://127.0.0.1:8001/mcp                                                                              
INFO:     Started server process [26956]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://127.0.0.1:8001 (Press CTRL+C to quit)

It will listen on http://localhost:8001/mcp.

---

🔗 Step 4: Connect via AI Toolkit in VS Code

1. Open VS Code with the AI Toolkit extension installed.
2. Go to AI Toolkit → MCP Servers → Add Server.
3. Choose HTTP transport.
4. Enter:
   • Name: Temp Conversion MCP
   • URL: http://localhost:8001/mcp
5. Save and connect.

Now, your AI Toolkit can call celsius_to_fahrenheit and fahrenheit_to_celsius directly as MCP tools.

---

✅ Step 5: Test with VS Code > mcp.json

• Add .vscode\mcp.json file at the root of the project and add following content in it.

{
  "servers": {
    "tempconversion-mcp-http": {
        "type": "http",
        "url": "http://127.0.0.1:8001/mcp"            
    },
    "tempconversion-mcp-stdio": {
      "type": "stdio",
      "command": "uv",
      "cwd": "${workspaceFolder}",
      "args": [
        "run",
        "tempconversion-mcp-stdio.py"
      ]
    },
    "tempconversion-mcp-stdio-debug": {
      "type": "stdio",
      "command": "uv",
      "cwd": "${workspaceFolder}",
      "args": [
        "run",
        "--",
        "python",
        "-m",
        "debugpy",
        "--listen",
        "0.0.0.0:5678",
        "tempconversion-mcp-stdio.py"
      ]
    }
  }
}

• Use MCP tools command from mcp.json to start / stop / restart fastmcp server

✅ Step 6: Test in AI Toolkit

• Open the AI Toolkit chat.
• Type:

  Use the Temp Conversion MCP server to convert 100 Celsius to Fahrenheit.
  

• The tool should respond with 212.0.

---

let’s add a manifest file (mcp.json) so your FastMCP server can be auto-discovered by AI Toolkit in VS Code.

---

📄 Step 6: Create mcp.json

In your project root (fastmcp-temp-server), add a file named mcp.json:

{
  "name": "Temp Conversion MCP",
  "version": "1.0.0",
  "description": "A simple FastMCP server for temperature conversion tools.",
  "transport": {
    "type": "http",
    "url": "http://localhost:8001"
  },
  "tools": [
    {
      "name": "celsius_to_fahrenheit",
      "description": "Convert Celsius to Fahrenheit",
      "input_schema": {
        "type": "object",
        "properties": {
          "celsius": { "type": "number" }
        },
        "required": ["celsius"]
      },
      "output_schema": {
        "type": "number"
      }
    },
    {
      "name": "fahrenheit_to_celsius",
      "description": "Convert Fahrenheit to Celsius",
      "input_schema": {
        "type": "object",
        "properties": {
          "fahrenheit": { "type": "number" }
        },
        "required": ["fahrenheit"]
      },
      "output_schema": {
        "type": "number"
      }
    }
  ]
}

---

📂 Step 8: Place the Manifest

• Save mcp.json in the same folder as your server.py.
• AI Toolkit will look for this manifest when you add the server.

---

🔗 Step 9: Connect in AI Toolkit

1. Open VS Code → AI Toolkit.
2. Go to MCP Servers → Add Server → Import from Manifest.
3. Select your mcp.json.
4. AI Toolkit will auto-register the server with the tools defined.

---

✅ Step 10: Test

• In AI Toolkit chat, try:

  Convert 0 Celsius to Fahrenheit using Temp Conversion MCP.
  

• It should return 32.0.

🔗 References

FastMCP

FastMCP Quickstart

How to Create an MCP Server in Python

FastMCP Cloud