Rack

What it does

Rack transforms your local environment into a data engineering pipeline. It connects APIs, databases, and Markdown notes (like Obsidian) into a queryable system powered by DuckDB. Instead of managing complex infrastructure, you describe what you want to track, and Claude writes the fetch, ingest, and derive scripts to keep your data fresh.

Tools

• rack init: Initializes a new data rack for a specific project.
• rack refresh: Executes the full pipeline: fetching raw data, ingesting it into DuckDB, and deriving Markdown dashboards.
• rack query: Allows executing direct SQL queries against the DuckDB database.
• rack lineage: Visualizes the data flow from source to final dashboard.
• rack signals: Reports notable conditions or alerts triggered by derive scripts.
• rack snapshot: Creates point-in-time copies of the database for historical analysis.

Installation

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "rack": {
      "command": "rack",
      "args": ["mcp", "serve"]
    }
  }
}

Supported hosts

• Claude Desktop
• Cursor
• VS Code Copilot