MSSQL MCP Server
# MSSQL MCP Server
[](https://opensource.org/licenses/MIT)
[](https://www.typescriptlang.org/)
[](https://nodejs.org/)
> **A Model Context Protocol (MCP) server that enables AI assistants to interact with both local SQL Server and Azure SQL Database through natural language.**
<a href="https://glama.ai/mcp/servers/@Nirmal123K/mssql-mcp">
<img width="380" height="200" src="https://glama.ai/mcp/servers/@Nirmal123K/mssql-mcp/badge" alt="MSSQL Server MCP server" />
</a>
## 🚀 What Makes This Special?
This MCP server **solves a critical limitation** of existing solutions by supporting **both local SQL Server and Azure SQL Database** connections. While the original Azure-Samples implementation only works with Azure SQL Database, this enhanced version enables:
- **🏠 Local Development**: Connect to local SQL Server instances using username/password authentication
- **☁️ Azure Cloud**: Full Azure SQL Database support with Azure AD authentication
- **🔄 Seamless Switching**: Use the same tools for both environments
**Quick Example:**
```text
You: "Show me all customers from New York"
AI: *securely queries your database and returns results in plain English*
You: "Create a table for storing product reviews"
AI: *generates and executes the appropriate CREATE TABLE statement*
```
## 📦 Quick Start
### Prerequisites
- **Node.js 18+**
- **SQL Server** (local) or **Azure SQL Database**
- **AI Assistant**: Claude Desktop or VS Code Agent
### Installation
```bash
git clone https://github.com/Nirmal123K/mssql-mcp.git
cd mssql-mcp-server
npm install && npm run build
```
### Configuration
**Local SQL Server:**
```bash
export SERVER_NAME="localhost"
export DATABASE_NAME="your_database"
export SQL_USER="your_username"
export SQL_PASSWORD="your_password"
export TRUST_SERVER_CERTIFICATE="true"
```
**Azure SQL Database:**
```bash
export SERVER_NAME="your-server.database.windows.net"
export DATABASE_NAME="your_database"
# Uses Azure AD authentication (run 'az login' first)
```
## ⚙️ AI Assistant Setup
### Claude Desktop
Add to `claude_desktop_config.json`:
**Local SQL Server:**
```json
{
"mcpServers": {
"mssql": {
"command": "node",
"args": ["/path/to/mssql-mcp-server/dist/index.js"],
"env": {
"SERVER_NAME": "localhost",
"DATABASE_NAME": "your_database",
"SQL_USER": "your_username",
"SQL_PASSWORD": "your_password",
"TRUST_SERVER_CERTIFICATE": "true"
}
}
}
}
```
**Azure SQL Database:**
```json
{
"mcpServers": {
"mssql": {
"command": "node",
"args": ["/path/to/mssql-mcp-server/dist/index.js"],
"env": {
"SERVER_NAME": "your-server.database.windows.net",
"DATABASE_NAME": "your_database",
"READONLY": "true"
}
}
}
}
```
### VS Code Agent
Create `.vscode/mcp.json` with similar configuration using `"type": "stdio"`.
**Config Locations:**
- **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
## 📋 Environment Variables
| Variable | Required | Description | Example |
|----------|----------|-------------|---------|
| `SERVER_NAME` | ✅ | SQL Server hostname | `localhost` or `server.database.windows.net` |
| `DATABASE_NAME` | ✅ | Target database name | `MyDatabase` |
| `SQL_USER` | 🔄 | Username (local SQL Server only) | `sa` or `myuser` |
| `SQL_PASSWORD` | 🔄 | Password (local SQL Server only) | `MyPassword123!` |
| `READONLY` | ❌ | Restrict to read-only operations | `true` or `false` (default) |
| `TRUST_SERVER_CERTIFICATE` | ❌ | Trust self-signed certificates | `true` or `false` (default) |
> **🔄 = Required for local SQL Server, not needed for Azure SQL Database**
## ✅ Verification
```bash
# Test the MCP server
echo '{"jsonrpc": "2.0", "id": 1, "method": "tools/list"}' | node dist/index.js
```
## 🛠️ Available Tools
| Tool | Purpose | Example |
|------|---------|---------|
| **`read_data`** | Query data with SELECT | "Show customers from New York" |
| **`insert_data`** | Add new records | "Insert new product with price $99" |
| **`update_data`** | Modify existing data | "Update order status to shipped" |
| **`create_table`** | Create new tables | "Create table for reviews" |
| **`create_index`** | Add database indexes | "Create index on customer email" |
| **`drop_table`** | Remove tables | "Drop temporary table" |
| **`list_tables`** | Show all tables | "What tables exist?" |
| **`describe_table`** | Show table structure | "Describe customers table" |
**Security Features:**
- ✅ SQL injection prevention
- ✅ Query validation and sanitization
- ✅ Required WHERE clauses for updates
- ✅ Read-only mode support
## 📚 Documentation
For detailed information, see our comprehensive guides:
- **[Usage Examples](docs/USAGE_EXAMPLES.md)** - Practical examples and real-world scenarios
- **[Advanced Configuration](docs/ADVANCED_CONFIGURATION.md)** - Complex setups and enterprise deployment
- **[Best Practices](docs/BEST_PRACTICES.md)** - Performance optimization and security guidance
- **[Troubleshooting](docs/TROUBLESHOOTING.md)** - Common issues and solutions
- **[Platform Notes](docs/PLATFORM_NOTES.md)** - Windows, macOS, Linux, and Docker specifics
## � Credits and Attribution
### Original Work
This project is a **fork** of the [Azure-Samples/SQL-AI-samples](https://github.com/Azure-Samples/SQL-AI-samples) repository, specifically building upon the **MssqlMcp** implementation created by Microsoft and the Azure team.
- **Original Repository**: [Azure-Samples/SQL-AI-samples](https://github.com/Azure-Samples/SQL-AI-samples)
- **Original Authors**: Microsoft Corporation and the Azure team
- **Original License**: MIT License
- **Original Focus**: Azure SQL Database with Azure AD authentication
### Key Enhancement
This fork **extends the original Azure-only MCP server** to support **local SQL Server databases** with username/password authentication, making it accessible for:
- 🏠 **Local Development**: Connect to local SQL Server instances
- 🏢 **On-Premises Deployments**: Support enterprise SQL Server installations
- 🔄 **Hybrid Environments**: Seamlessly switch between local and Azure databases
- 🚀 **Broader Accessibility**: Remove Azure dependency for local development
### What We Added
| Feature | Original Repository | This Fork |
|---------|-------------------|-----------|
| **Azure SQL Database** | ✅ Azure AD only | ✅ Azure AD support |
| **Local SQL Server** | ❌ Not supported | ✅ Username/password auth |
| **Authentication Methods** | 1 (Azure AD) | 3 (Azure AD, SQL Server, Windows) |
| **Target Audience** | Azure users only | Local + Azure + On-premises |
| **Development Setup** | Requires Azure account | Works with local SQL Server |
### Acknowledgments
We are **deeply grateful** to Microsoft and the Azure team for creating the foundational architecture, security implementation, and tool design that made this enhanced version possible.
**This enhanced version would not exist without their excellent foundational work.**
---
> **License**: MIT License - see [LICENSE](LICENSE) file for details.TDQS
Scored across 8 tools
Each tool has a clearly distinct purpose with no ambiguity: create_index (index creation), create_table (table creation), describe_table (schema inspection), drop_table (table deletion), insert_data (data insertion), list_table (table listing), read_data (data querying), and update_data (data modification). The tools cover different aspects of database operations without overlap.
All tools follow a consistent verb_noun pattern (e.g., create_table, describe_table, drop_table, insert_data, list_table, read_data, update_data). The naming is uniform and predictable, making it easy for agents to understand the action and target resource.
With 8 tools, this server is well-scoped for MSSQL database operations. Each tool serves a specific and necessary function in the domain, such as table management, data manipulation, and schema inspection, without being overly sparse or bloated.
The tool set provides strong coverage for core database operations, including table lifecycle (create, describe, list, drop), data CRUD (insert, read, update), and index creation. A minor gap exists with no explicit tool for deleting data (e.g., delete_data) or managing indexes beyond creation, but agents can work around this using update_data or other methods.