MCP Server ODBC via SQLAlchemy

A lightweight MCP (Model Context Protocol) server for ODBC built with FastAPI, pyodbc, and SQLAlchemy. This server is compatible with Virtuoso DBMS and other DBMS backends that implement a SQLAlchemy provider.

Features

• Get Schemas: Fetch and list all schema names from the connected database.
• Get Tables: Retrieve table information for specific schemas or all schemas.
• Describe Table: Generate a detailed description of table structures, including:
  • Column names and data types
  • Nullable attributes
  • Primary and foreign keys
• Search Tables: Filter and retrieve tables based on name substrings.
• Execute Stored Procedures: In the case of Virtuoso, execute stored procedures and retrieve results.
• Execute Queries:
  • JSONL result format: Optimized for structured responses.
  • Markdown table format: Ideal for reporting and visualization.

Prerequisites

1. Install uv:

   pip install uv
   

   Or use Homebrew:

   brew install uv
   

2. unixODBC Runtime Environment Checks:

3. Check installation configuration (i.e., location of key INI files) by running: odbcinst -j

4. List available data source names by running: odbcinst -q -s

5. ODBC DSN Setup: Configure your ODBC Data Source Name (~/.odbc.ini) for the target database. Example for Virtuoso DBMS:

   [VOS]
   Description = OpenLink Virtuoso
   Driver = /path/to/virtodbcu_r.so
   Database = Demo
   Address = localhost:1111
   WideAsUTF16 = Yes
   

6. SQLAlchemy URL Binding: Use the format:

   virtuoso+pyodbc://user:password@VOS
   

Installation

Clone this repository:

git clone https://github.com/OpenLinkSoftware/mcp-sqlalchemy-server.git
cd mcp-sqlalchemy-server

Environment Variables

Update your .envby overriding the defaults to match your preferences

ODBC_DSN=VOS
ODBC_USER=dba
ODBC_PASSWORD=dba
API_KEY=xxx

Configuration

For Claude Desktop users: Add the following to claude_desktop_config.json:

{
  "mcpServers": {
    "my_database": {
      "command": "uv",
      "args": ["--directory", "/path/to/mcp-sqlalchemy-server", "run", "mcp-sqlalchemy-server"],
      "env": {
        "ODBC_DSN": "dsn_name",
        "ODBC_USER": "username",
        "ODBC_PASSWORD": "password",
        "API_KEY": "sk-xxx"
      }
    }
  }
}

Usage

Database Management System (DBMS) Connection URLs

Here are the pyodbc URL examples for connecting to DBMS systems that have been tested using this mcp-server.

Tools Provided

Troubleshooting

For easier troubleshooting:

1. Install the MCP Inspector:

   npm install -g @modelcontextprotocol/inspector
   

2. Start the inspector:

   npx @modelcontextprotocol/inspector uv --directory /path/to/mcp-sqlalchemy-server run mcp-sqlalchemy-server
   

Access the provided URL to troubleshoot server interactions.