MCP SQL Server

A Model Context Protocol (MCP) server that exposes a SQL Server database as tools for AI assistants like Claude, Gemini, and Copilot.

Features

• test_connection — Verify database connectivity
• list_tables — List all tables in the database
• query_sql — Execute arbitrary SQL queries (read/write with 100-row limit)
• get_oee_by_line — Calculate OEE metrics (Availability, Performance, Quality) for a production line
• get_downtime_analysis — Aggregate downtime reasons sorted by duration
• get_shift_performance — KPI analytics for a specific shift and date
• get_live_kpi — Fetch the latest OEE value from the live table

Prerequisites

• Python 3.10+
• Microsoft ODBC Driver 17+ for SQL Server
• Access to a SQL Server database

Installation

# Clone the repository
git clone https://github.com/gigimento/mcp-sql-server.git
cd mcp-sql-server

# Create virtual environment
python -m venv .venv
.venv\Scripts\activate  # Windows
# source .venv/bin/activate  # Linux/macOS

# Install dependencies
pip install -r requirements.txt

Configuration

Copy .env.example to .env and adjust:

DB_SERVER=localhost\SQLEXPRESS
DB_NAME=YourDatabase
DB_DRIVER=ODBC Driver 17 for SQL Server
DB_TRUSTED_CONNECTION=yes

Or set the corresponding environment variables directly.

Usage

With Claude Desktop / OpenCode

Add to your MCP config:

{
  "mcpServers": {
    "sql-server": {
      "command": "python",
      "args": ["path/to/server.py"],
      "env": {
        "DB_SERVER": "localhost\\SQLEXPRESS",
        "DB_NAME": "YourDatabase"
      }
    }
  }
}

Standalone

python server.py

The server communicates over stdio using the MCP protocol.

Expected Database Schema

The OEE tools expect tables similar to:

• ProductionData — columns: Line, Date, Shift, Runtime, PlannedTime, GoodQty, RejectQty, TotalQty, PlannedQty
• Downtime — columns: Line, Date, DowntimeReason, Duration
• OEE_Live — columns: Line, OEE, Timestamp

You can adapt the queries to match your schema by editing the tool functions in server.py.

License

MIT