This hub covers how to configure Claude Code for dbt projects and use it for testing, documentation, and debugging — six notes covering systematic, repeatable workflows. Assumes familiarity with Claude Code basics and dbt on BigQuery.
Prerequisites
- Claude Code installed and functional
- A dbt project targeting BigQuery
- Familiarity with basic Claude Code usage (prompting, file editing, command execution)
Reading Order
Foundation
-
CLAUDE.md as Project Memory — How to give Claude persistent project context. Covers the reactive approach to building CLAUDE.md that converges on instructions that actually matter.
-
Claude Code Hooks — Deterministic guardrails that enforce rules regardless of prompting: auto-format SQL, block edits to production files, trigger compilation after changes.
-
Claude Code Slash Commands for dbt — Reusable, team-shareable commands for common workflows; turns multi-step prompts into single invocations.
Workflows
-
TDD with Claude Code for dbt — Five-step test-driven development pattern: write tests first, let the agent iterate to green, refactor.
-
dbt Documentation with Claude Code — The codegen-plus-AI pattern for model documentation, docs blocks for cross-model consistency, and Mermaid lineage diagrams.
-
Debugging dbt with Claude Code — Letting Claude face errors directly rather than pre-gathering evidence; subagents for complex investigations that need fresh context.