Azure SQL MCP Server

Azure SQL MCP Server — Complete Guide

A Model Context Protocol (MCP) server that connects Microsoft Copilot Studio to your Azure SQL Database. Supports 12 tools for querying, CRUD operations, schema inspection, search, and chart visualization.

Architecture

1. Prerequisites

• Python 3.8+
• ODBC Driver 18 for SQL Server — Download here
• Azure SQL Database with server hostname, database name, username, and password
• Cloudflare Tunnel (for local testing) — winget install Cloudflare.cloudflared

Install ODBC Driver

Windows: Download and install from the link above.

macOS:

brew tap microsoft/mssql-release https://github.com/Microsoft/homebrew-mssql-release
brew update
brew install msodbcsql18 mssql-tools18

Linux (Ubuntu/Debian):

curl https://packages.microsoft.com/keys/microsoft.asc | sudo apt-key add -
curl https://packages.microsoft.com/config/ubuntu/$(lsb_release -rs)/prod.list | sudo tee /etc/apt/sources.list.d/mssql-release.list
sudo apt-get update
sudo ACCEPT_EULA=Y apt-get install -y msodbcsql18

2. Installation

git clone <your-repo-url>
cd azure-sql-mcp-server
python -m venv venv
venv\Scripts\activate          # Windows
# source venv/bin/activate     # macOS/Linux
pip install -r requirements.txt

3. Configuration

Create a .env file in the project root:

AZURE_SQL_SERVER=your-server.database.windows.net
AZURE_SQL_DATABASE=your-database-name
AZURE_SQL_USERNAME=your-username
AZURE_SQL_PASSWORD=your-password
AZURE_SQL_DRIVER=ODBC Driver 18 for SQL Server

Never commit .env to version control. Add it to .gitignore.

4. Code Fixes

The MCP SDK requires specific configuration. Apply these 3 fixes to azure_sql_mcp.py:

Fix 1: Lifespan function signature

# ❌ BEFORE
@asynccontextmanager
async def app_lifespan():

# ✅ AFTER — FastMCP passes the server instance
@asynccontextmanager
async def app_lifespan(server: FastMCP):

Fix 2: Host and port on the constructor

# ❌ BEFORE
mcp = FastMCP("azure_sql_mcp", lifespan=app_lifespan)

# ✅ AFTER — host/port go on the constructor, NOT on run()
mcp = FastMCP("azure_sql_mcp", host="0.0.0.0", port=8000, lifespan=app_lifespan)

Fix 3: HTTP transport

# ❌ BEFORE
if __name__ == "__main__":
    mcp.run()

# ✅ AFTER — streamable-http (with hyphen) for Copilot Studio
if __name__ == "__main__":
    mcp.run(transport="streamable-http")

5. Running the Server

cd azure-sql-mcp-server
.\venv\Scripts\Activate.ps1       # Windows
python azure_sql_mcp.py

You should see:

INFO: Initializing Azure SQL MCP server...
INFO: Database connection established
INFO: Database connection verified
INFO: Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)

MCP endpoint: http://localhost:8000/mcp

6. Exposing to the Internet

Copilot Studio needs a public HTTPS URL. Use Cloudflare Tunnel (free, no signup).

Why not ngrok? ngrok's free tier shows a browser warning page (ERR_NGROK_6024) that blocks API clients like Copilot Studio.

Steps

1. Install (one-time):

   winget install Cloudflare.cloudflared
   

   Close and reopen your terminal after install.

2. In a new terminal (keep the server running in the first one):

   cloudflared tunnel --url http://localhost:8000
   

3. Copy the URL from the output:

   https://electronic-annie-jose-spoken.trycloudflare.com
   

4. Your MCP server URL for Copilot Studio:

   https://electronic-annie-jose-spoken.trycloudflare.com/mcp
   

URLs change on restart. For a permanent URL, deploy to Azure App Service (see Production Deployment).

7. Copilot Studio Setup

Step 1: Add the MCP Server

1. Go to Copilot Studio
2. Open your agent → Tools → Add a tool → New tool → Model Context Protocol
3. Fill in: