How to Connect AI Assistants to SQL Server with MCP: Step-by-Step Guide

Unlock powerful AI-driven data exploration in your Microsoft SQL Server environment with the Model Context Protocol (MCP). This practical guide walks backend engineers and API developers through securely connecting AI assistants like Claude to SQL Server, enabling advanced querying and workflow automation―all without compromising database security.

💡 Before configuring SQL Server MCP, consider Apidog for your API design, testing, and documentation needs. Apidog streamlines API integration, provides structured models, and enhances team collaboration. For advanced automation, Apidog pairs seamlessly with Zapier MCP.

button

What Is MS SQL Server MCP?

MS SQL Server MCP is a dedicated server that implements the Model Context Protocol (MCP), acting as a secure gateway between your Microsoft SQL Server database and AI-powered tools. Instead of exposing your database directly to external AI models—a major security risk—MCP provides a controlled, auditable interface for:

Listing database tables
Reading table data securely
Executing SQL queries with robust error handling

All access is mediated via environment variables, with detailed logging and structured error management to help maintain data integrity and compliance.

Key Features of MS SQL Server MCP

The MS SQL Server MCP server is built for professional database operations:

Resource Discovery: List tables as accessible resources
Secure Table Reads: Query specific table contents safely
Custom SQL Execution: Run parameterized SQL with comprehensive error handling
Environment-Based Security: Manage credentials via environment variables, never in code
Extensive Logging: Track all queries and access attempts
Automated Dependency Management: System libraries like FreeTDS are installed automatically

For API teams seeking structured, secure, and automated workflows, pairing MCP with Apidog accelerates integration and documentation.

How to Use VSCode MCP Server—A Full Tutorial

Installing MS SQL Server MCP Server

Getting started with the MS SQL Server MCP server is straightforward:

1. Install the MCP Server Package

Open your terminal and run:

pip install mssql-mcp-server

How It Works:

pip downloads and installs the mssql-mcp-server package from PyPI
All required system dependencies (e.g., FreeTDS for SQL Server communication) are handled automatically

For more MCP server options, see HiMCP.ai - Discover 1682+ MCP Servers.

Connecting MCP Server with Claude Desktop

Once installed, you can link the MS SQL Server MCP server to AI assistants like Claude Desktop.

1. Update Claude Desktop Configuration

Open your claude_desktop_config.json (usually found under Developer or Advanced settings) and add:

{
  "mcpServers": {
    "mssql": {
      "command": "uv",
      "args": [
        "--directory",
        "path/to/mssql_mcp_server",
        "run",
        "mssql_mcp_server"
      ],
      "env": {
        "MSSQL_SERVER": "localhost",
        "MSSQL_USER": "your_username",
        "MSSQL_PASSWORD": "your_password",
        "MSSQL_DATABASE": "your_database"
      }
    }
  }
}

Implementation Tips:

Replace "path/to/mssql_mcp_server" with your actual installation path. To find this, run:
```
import mssql_mcp_server
print(mssql_mcp_server.__file__)
```
Substitute database credentials as appropriate for your environment.
"uv" is used as the process runner; ensure it's installed.

2. Restart Claude Desktop

After updating the config, completely close and reopen Claude Desktop for changes to take effect.

Testing Your MCP Connection in Claude

With configuration complete, you can now interact with SQL Server via Claude:

1. List Database Tables

In Claude Desktop, type:

List the tables in my SQL Server database

Claude should return a full list of available tables if the connection is successful.

list tables

2. Query Table Contents

Ask Claude:

Give me a description of each table based on their contents

Claude will analyze and describe your database schema and sample data.

tables description

3. Visualize Table Data in Chat

You can prompt Claude to create chats that visualize table data directly in the conversation—ideal for quick team overviews or QA reviews.

table data visualisation

Running MCP Server Standalone

You’re not limited to Claude Desktop. The MS SQL Server MCP server can run independently for broader integration.

Steps:

Clone the Repository:

git clone https://github.com/RichardHan/mssql_mcp_server.git
cd mssql_mcp_server

Install Dependencies:
```
pip install -r requirements.txt
```
Start the Server:
```
python -m mssql_mcp_server
```
The server listens on a default port (usually 5000), ready for connections from any MCP-compatible application.

Development & Contribution Guide

Interested in extending or contributing to the MCP project? Here’s how to set up a dev environment:

Clone the Repo:

git clone https://github.com/RichardHan/mssql_mcp_server.git
cd mssql_mcp_server

Create a Virtual Environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

Install Dev Dependencies:
```
pip install -r requirements-dev.txt
```
Run Tests:
```
pytest
```
Run all or specific tests to validate your changes.

Conclusion

With MS SQL Server MCP, you can securely connect AI assistants like Claude to your SQL Server databases—enabling advanced querying, data exploration, and automation without sacrificing security. MCP provides a standardized, auditable interface for AI-driven workflows, making it a strong fit for professional API teams and backend engineers seeking to integrate intelligent assistants into their data stack.

Integrate MCP with Apidog to further streamline your API workflows, from design to documentation and testing.

button

Apidog all in one image