Recipes
Recipes are end-to-end task guides. Each one starts from a user goal, shows which interface to use, and walks through the commands or UI steps to reach a clear outcome.
Choose a recipe
| Goal | Recipe |
|---|---|
| Check whether a dbt run succeeded and identify failed nodes | Debug a failed run |
| Rank slow models and find timing bottlenecks | Investigate slow models |
| Understand which models would break if you changed one | Find model impact |
| Produce a dbt health summary in GitHub Actions or CI | Generate CI health summary |
| Move from CLI JSON output to visual browser investigation | Open CLI result in Web |
| Let an AI agent query dbt artifacts safely | Ask an agent about a dbt run |
Prerequisites for all recipes
manifest.jsonandrun_results.jsonunder a dbt target directory- Node.js 20+
Run npx @dbt-tools/cli status --dbt-target ./target --json first to confirm artifacts are readable.
No real dbt project? Use Try with a sample project to generate artifacts with a public sample repo, then run any recipe against
./target.
How recipes relate to existing workflows
The site also has detailed Workflows pages under each interface section (CLI, Web, Agents). Recipes and workflows cover overlapping ground from different angles:
- Recipes start with a user goal and route to the right interface.
- Workflows start with a specific interface and walk through its steps.
Both are valid. Use whichever framing matches how you think about your task.
Related
- Choose by goal — quick routing table
- 5-minute quickstart — get your first command running
- Deploy — local, S3, GCS, and CI configuration
- Trust & Safety — data boundaries, agent safety, and licensing