Skip to main content
Community Hub

Query Lakehouse with AI agents

TL;DR

Query Lakehouse metadata using natural language in AI coding agents like Claude Code. Install the atlan-lakehouse skill, which detects your platform and generates appropriate SQL queries.

Your AI can read this via Docs MCPcurl -fsSL "https://docs.atlan.com/install-docs-mcp" | bashConnect

The atlan-lakehouse skill enables natural language queries on Lakehouse metadata in AI agents like Claude Code. The skill detects your platform (Snowflake via Cortex Code, Databricks via Genie Code, or Python via PyIceberg) and generates SQL automatically, letting you explore your catalog metadata without writing SQL manually.

Prerequisites​

Before you begin, make sure you have:

Install Agent skill​

  1. Clone the Lakehouse Solutions repository to your machine:

    git clone https://github.com/atlanhq/lakehouse-solutions.git

  2. Create the Claude Code skills directory if it doesn't already exist:

    mkdir -p ~/.claude/skills

  3. Symlink the skill into your Claude Code skills directory:

    ln -s /path/to/lakehouse-solutions/skills/atlan-lakehouse ~/.claude/skills/atlan-lakehouse

    Replace /path/to/lakehouse-solutions with the actual path where you cloned the repository.

Query Lakehouse using natural language​

Once the skill is installed, Claude Code activates it automatically when you ask Lakehouse-related questions.

  1. Open Claude Code in your terminal:

    claude

  2. Ask a question in natural language. For example:

    • "What percentage of my Snowflake tables have descriptions?"
    • "Which assets are missing owners?"
    • "Show me lineage gaps in the sales domain."

    The skill identifies your platform and runs the appropriate SQL query in your Lakehouse.

See also​