// the find
dbt-labs/dbt-project-evaluator
This package contains macros and models to find DAG issues automatically
A dbt package that audits your dbt project against dbt Labs' own conventions — DAG structure, test coverage, documentation, naming, materializations, and governance. It works by running as a dbt package itself, so violations show up as failed dbt tests in your normal CI run. For any team that has let a dbt project grow organically and suspects it's become a mess.
The approach of using dbt to audit dbt is genuinely clever — you get rule violations as failed tests in the same pipeline you're already running, no external tooling required. The `int_all_dag_relationships` model is a real bonus: a queryable, tabular representation of your DAG that's useful well beyond just this package. Adapter support is broad (BigQuery, Snowflake, Databricks, DuckDB, Redshift, Postgres, Trino, Athena, Fabric) and actually tested, not just listed. The exceptions mechanism via a seed CSV means you can acknowledge known violations without polluting CI with noise you can't fix today.
The rules are dbt Labs' opinionated conventions, not universal best practices — teams with legitimate deviations (say, staging models that intentionally join two sources) will spend meaningful time configuring exceptions rather than getting value. Running this adds non-trivial compute time to your dbt run since it materializes a lot of models against your warehouse; on large projects with slow warehouses this adds up. The `dispatch` override of core dbt macros for some adapters is a footgun — it can silently break if you're also using other packages that do the same thing. Documentation coverage checks are binary (documented or not) and don't catch low-quality descriptions, so you can pass with placeholder text.