Discover MSSQL MCP Server (Preview): A New Era of Conversing with Your Data

Imagine interacting with your data not through cryptic SQL queries, but through straightforward, natural conversations. For years, tapping into the true capabilities of SQL Server databases has required specialized skills—reserved mostly for developers and DBAs who can navigate complex query languages and tools. But what if anyone, regardless of their technical background, could simply ask a database a question and receive a meaningful, actionable reply?

We’re excited to introduce the MSSQL MCP Server, now available in preview as an open-source project. This groundbreaking technology redefines how you work with your SQL Server databases by opening them up directly to AI agents, bridging the gap between raw data and intuitive interaction.

What is MCP?

At the heart of this shift is the Model Context Protocol (MCP)—an open standard built to facilitate smooth collaboration between AI agents and external systems. MCP introduces two key components:

• MCP Clients, embedded in AI-driven applications.
• MCP Servers, which expose a system’s capabilities.

MCP transforms how AI integrates with various tools and data sources. Instead of building countless custom connectors, developers can implement a single MCP server, and any MCP client can communicate with it. This “write once, integrate everywhere” approach simplifies what used to be a tangle of integrations.

An MCP Server securely shares select capabilities—termed as “tools”, “resources”, and “prompts”—which AI models can discover, understand, and trigger using natural language. This standardized layer lets AI interact with complex back-end systems safely and efficiently.

👉 To explore more, check out the official MCP protocol site and the GitHub organization.

MSSQL MCP Server: Your AI-Powered Gateway to SQL Server

Building on the MCP framework, the MSSQL MCP Server connects directly to your SQL Server environments, providing a reliable conduit for AI assistants. It’s available in both .NET and Node.js versions, giving you flexibility depending on your stack.

In this guide, we’ll focus on the Node.js implementation, though you’ll find detailed .NET instructions in the GitHub repository.

With this powerful server in place, you can seamlessly connect to:

• On-premises SQL Server
• Azure SQL Database
• SQL Database in Microsoft Fabric

Switching between these is as simple as updating a connection string. Whether it’s schema management, data manipulation, or running queries, you’ll now do it through natural language—thanks to this robust architecture.

MSSQL MCP in Practice

Tools at Your Fingertips

The MSSQL MCP Server comes with a full suite of tools that enable AI to perform various database operations:

Tool

What It Does

List Tables

Lists all tables in the database, helping the AI understand the data landscape.

Describe Table

Retrieves schema details, column info, and data types for a specific table.

Create Table

Lets the AI define and create new tables, managing columns and types via simple instructions.

Drop Table

Safely deletes tables—ideal for cleaning up or iterating on schema changes.

Insert Data

Inserts rows into tables, including batch operations for high efficiency.

Read Data

Queries and retrieves records with conditions and sorting, replacing manual SELECT statements.

Update Data

Updates existing records based on conditions, handling maintenance or corrections.

Create Index

Builds indexes on chosen columns to optimize query performance.

Integrating with MCP Clients

The real power of MSSQL MCP Server shines when you pair it with AI-enhanced environments like Visual Studio Code or conversational platforms like Claude Desktop.

Getting Started with Node.js MSSQL MCP Server

1. Prerequisites

• Install Node.js and Git.

2. Clone the Repository

bash

git clone https://github.com/Azure-Samples/SQL-AI-samples.git

3. Navigate to the Node Project

bash

cd SQL-AI-samples/MssqlMcp/Node

4. Install and Build

bash

npm install

Typically, npm install triggers any build steps defined in package.json. If needed, consult the file for specifics.

5. Locate Your Entry File

After the build completes, find index.js inside the dist directory. Copy its full path—you’ll need it when configuring your client.

Using with Visual Studio Code (GitHub Copilot)

VS Code, combined with GitHub Copilot Chat, offers out-of-the-box support for MCP servers.

1. Install VS Code and GitHub Copilot.
2. Configure the MCP Server in settings.json:

json

{
  "mcp": {
    "servers": {
      "MSSQL MCP": {
        "type": "stdio",
        "command": "node",
        "args": ["C:\\path\\to\\dist\\index.js"],
        "env": {
          "SERVER_NAME": "demo-server.database.windows.net",
          "DATABASE_NAME": "db-events",
          "READONLY": "false"
        }
      }
    }
  }
}        

Replace paths and values with your actual setup.

1. Enable Agent Mode: Switch to Agent mode in VS Code and select “MSSQL MCP” from the tools list.
2. Start the Server: Use the Start button at the top of your settings.json.
3. Talk to Your Database: Use natural prompts in the chat, such as:

“Which database am I connected to?”

The first time, you’ll authenticate using Entra ID, so make sure your database access is set up accordingly.

Using with Claude Desktop

Claude Desktop also supports MCP connections, letting you chat directly with your database.

1. Install Claude Desktop.
2. Edit Configuration: Navigate to File > Settings > Developer > Edit Config, opening claude_desktop_config.json.

json

{
  "mcpServers": {
    "MSSQL MCP": {
      "type": "stdio",
      "command": "node",
      "args": ["C:\\path\\to\\dist\\index.js"],
      "env": {
        "SERVER_NAME": "demo-server.database.windows.net",
        "DATABASE_NAME": "db-events",
        "READONLY": "false"
      }
    }
  }
}        

Customize with your actual paths and connection details.

1. Restart Claude Desktop: Exit via the system tray, then relaunch from your Start Menu or Applications.
2. Chat with Your Data: Start using conversational prompts. Like VS Code, the first time will prompt Entra authentication.

What’s Next? Your First Natural Conversation with Data

With everything set up, you’re ready to experience data interaction like never before. No SQL syntax, just simple, human-friendly questions typed into your AI-enhanced editor.

Try:

“What tables exist in my database?”

Or

“Show me the first 10 rows from the events table.”

Wrapping Up

The MSSQL MCP Server (Preview) is more than just a new tool—it’s a paradigm shift. By harnessing the open Model Context Protocol, it democratizes access to powerful SQL Server capabilities, enabling anyone to converse with data naturally through AI agents.

This is just the start. In future articles, we’ll dive into hands-on scenarios, such as how event managers use this technology to streamline planning and reporting—turning intricate data tasks into simple chats.