dbt projects as a first-class Windmill runtimeImport an unmodified dbt project, the same directory you run
dbt builddbt build against locally, and run it on your own workers. A dbt_project.ymldbt_project.yml is all Windmill needs to pick it up.
The whole project deploys with the script, bundled and versioned atomically. Nothing is cloned at run time. Warehouses (postgres, snowflake, bigquery, databricks, redshift, and any adapter dbt supports) are configured once in workspace settings and referenced by name. Every model, seed, snapshot and source becomes a dbt://<warehouse>/<schema>/<name>dbt://<warehouse>/<schema>/<name> asset, and dbt's own ref() lineage is drawn as edges between them. Hit Refresh modelsRefresh models and Windmill runs a real dbt parsedbt parse to rebuild the graph from dbt's manifest. Native scripts can read those assets by URI, so a dbt build lands in the same lineage as the rest of your jobs.
Read more: https://www.windmill.dev/docs/getting_started/scripts_quickstart/dbt
