What is Mcp Mirror

mcp-mirror is a simple MCP server designed to mirror tables from ODBC data sources to SQLite databases, facilitating data transfer without loading large datasets into the context window of Claude Desktop.

Use cases

Use cases include migrating customer data from an ODBC database to SQLite for reporting purposes, synchronizing inventory data between systems, and enabling data analysis in environments where SQLite is preferred.

How to use

To use mcp-mirror, clone the repository, install the necessary dependencies using the UV package manager, configure the ODBC and SQLite database paths, and run the server. Once running, you can execute commands to mirror tables from ODBC to SQLite.

Key features

Key features include: single-purpose functionality for mirroring tables, compatibility with any ODBC-compatible database, automatic creation of SQLite tables with the appropriate schema, batch processing for large tables, preservation of data types during transfer, and special handling for ProvideX/Sage 100 connections.

Where to use

mcp-mirror can be used in data integration scenarios, database migration projects, and applications requiring data synchronization between ODBC sources and SQLite databases.

Mirror MCP Server

A simple MCP (Model Context Protocol) server that provides a tool for mirroring tables from ODBC to SQLite. This allows Claude Desktop to move data between databases without loading large datasets into its context window.

Features

• Single purpose: mirror tables from ODBC to SQLite
• Works with any ODBC-compatible database
• Automatically creates SQLite tables with appropriate schema
• Handles large tables with batch processing
• Preserves data types during transfer
• Special handling for ProvideX/Sage 100 connections

Prerequisites

• Python 3.10 or higher
• UV package manager
• ODBC drivers for your source database
• Claude Desktop with both ODBC and SQLite MCP servers configured

Installation

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

# Install with UV
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e .

Configuration

This server requires both an ODBC configuration file and a path to a SQLite database:

mirror-mcp-server --odbc-config /path/to/odbc-config.ini --sqlite-db /path/to/sqlite.db

Claude Desktop Integration

Add the Mirror MCP server to your Claude Desktop configuration:

{
  "mcpServers": {
    "mirror": {
      "command": "uv",
      "args": [
        "--directory",
        "C:\\path\\to\\mirror-mcp-server",
        "run",
        "mirror-mcp-server",
        "--odbc-config",
        "C:\\path\\to\\odbc-config.ini",
        "--sqlite-db",
        "C:\\path\\to\\sqlite.db"
      ]
    }
  }
}

Usage

Once the server is running and connected to Claude Desktop, you can use the following tools:

1. mirror-table

Copies a table from ODBC to SQLite:

# Example query in Claude
Mirror the Customer table from ODBC to SQLite

Parameters:

• source_table: Name of the source table in ODBC (required)
• dest_table: Name of the destination table in SQLite (optional, defaults to source name)
• connection_name: Name of the ODBC connection to use (optional, uses default if not specified)
• overwrite: Whether to overwrite existing data (optional, defaults to false)

2. list-odbc-connections

Lists all available ODBC connections from the config file:

# Example query in Claude
List the available ODBC connections

Example Workflows

1. Mirror a customer table and then analyze it:

   First, mirror the AR_Customer table from the ODBC database to SQLite.
   Then, show me the total number of customers by region from the SQLite database.
   

2. Create a reporting database:

   Mirror the following tables from ODBC to SQLite:
   - Sales_Header
   - Sales_Detail
   - Product
   - Customer
   
   Then create a sales summary report by month using the SQLite data.
   

Troubleshooting

• Connection errors: Verify that the ODBC configuration file is correct
• Table not found: Check that the table name is spelled correctly and accessible to the ODBC user
• Permission issues: Ensure the SQLite database path is writable

License

MIT License