Fetching Dbt Docs

What it does

This skill enables AI agents to retrieve and parse documentation from dbt projects, allowing them to understand model definitions, column descriptions, and relationship metadata without manual inspection. It bridges the gap between raw data pipelines and semantic understanding for automated analysis.

When to use it

• An agent needs to explain the lineage of a specific metric to a stakeholder.
• You want an AI to audit data quality rules defined in your dbt project files.
• An agent must generate SQL queries that respect existing table constraints and foreign keys.
• You need to quickly summarize the purpose of a complex transformation model before running it.

Key capabilities

• Fetches documentation from local or remote dbt projects.
• Parses metadata including models, sources, macros, and exposures.
• Extracts column-level descriptions and data types.
• Identifies relationships between tables based on defined references.

Example prompts

• "Explain the logic behind the customer_lifetime_value model using the project documentation."
• "List all columns in the orders table that have a description mentioning 'revenue'."
• "What are the upstream dependencies for the monthly_sales_summary model?"

Tips & gotchas

Ensure your dbt project has valid docs.yml files or descriptions enabled in dbt_project.yml, as the skill relies on this metadata. The agent cannot access database runtime data; it only reads static documentation artifacts generated by dbt.