MCP 服务器

MSSQL Database MCP Server

Enables AI assistants to securely interact with Microsoft SQL Server databases to query data, inspect schemas, and retrieve metadata with read-only operations by default and optional write capabilities.

README

MSSQL Database MCP Server

An enterprise-grade Model Context Protocol (MCP) server for Microsoft SQL Server. This server enables AI assistants (like Claude, Cursor, Windsurf) to securely interact with MSSQL databases to query data, inspect schemas, and retrieve metadata.

Features

Read-Only by Default: Strict safety controls to prevent accidental writes.
Schema Introspection: Tools to list databases, tables, and describe table schemas.
Secure Authentication: Supports SQL Server authentication with encryption.
Docker Ready: Easy deployment using Docker or Docker Compose.
Comprehensive Logging: Structured JSON logging for auditing.

Quick Start

1. Build the Docker Image

docker build -t mssql-mcp-server:latest .

2. Configure Environment

Create a .env file based on .env.example:

cp .env.example .env

Edit .env with your MSSQL connection details:

MSSQL_HOST=host.docker.internal
MSSQL_PORT=1433
MSSQL_DATABASE=master
MSSQL_USER=sa
MSSQL_PASSWORD=YourStrong@Passw0rd
MSSQL_ENCRYPT=true
MSSQL_TRUST_SERVER_CERTIFICATE=true

3. Run with Docker Compose

docker-compose up -d

Tools Available

The server exposes the following MCP tools:

Core Tools (Always Available)

mssql_query: Execute SELECT queries (with safety checks)
mssql_list_databases: List all accessible databases
mssql_list_tables: List tables in a specific database/schema
mssql_describe_table: Get detailed schema information (columns, PKs) for a table
mssql_pool_stats: Get connection pool statistics for monitoring

Advanced Tools (Require Write Operations Enabled)

⚠️ These tools require MSSQL_ALLOW_WRITE_OPERATIONS=true

mssql_execute_procedure: Execute stored procedures with parameters
mssql_execute_write: Execute INSERT, UPDATE, DELETE statements with transaction safety

Performance Features

Connection Pooling

The server implements intelligent connection pooling for optimal performance:

Reusable Connections: Connections are pooled and reused across queries
Thread-Safe: Safe for concurrent requests
Auto-Validation: Connections are validated before use
Configurable Limits: Control min/max pool size via environment variables
Connection Lifetime: Automatic rotation of long-lived connections
Monitoring: Track pool usage with mssql_pool_stats tool

Configuration:

MIN_POOL_SIZE=2          # Minimum connections to maintain
MAX_POOL_SIZE=10         # Maximum concurrent connections
IDLE_TIMEOUT=300         # Seconds before idle connection closes
CONNECTION_LIFETIME=1800 # Max lifetime of a connection (30 min)

Monitor Pool Health:

{
  "total_connections": 5,
  "available_connections": 3,
  "max_connections": 10,
  "min_connections": 2
}

IDE Integration

For detailed integration instructions for Claude Desktop, Cursor, and Windsurf, see IDE-INTEGRATION.md.

Example Configuration (Generic)

Add this to your IDE's MCP configuration file:

{
  "mcpServers": {
    "mssql": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "--network", "host",
        "-e", "MSSQL_HOST=localhost",
        "-e", "MSSQL_USER=sa",
        "-e", "MSSQL_PASSWORD=password",
        "mssql-mcp-server:latest"
      ]
    }
  }
}

Development & Testing

The project includes an automated test suite that spins up a test MSSQL container.

Run Tests

# Make scripts executable
chmod +x test-runner.sh

# Run full integration test suite
./test-runner.sh

Project Structure

src/: Source code
- server.py: Main MCP server entry point
- tools/: MCP tool implementations
- database/: Database connection and validation logic
tests/: Integration tests
docker/: Docker configuration

Security Notes

Write Operations: Disabled by default. To enable, set MSSQL_ALLOW_WRITE_OPERATIONS=true.
Credentials: Never commit .env files. Use Docker secrets in production.
Network: Use TLS (MSSQL_ENCRYPT=true) for non-local connections.