Load a dbt project as Dagster software-defined assets using @dbt_assets and DbtCliResource

Steps

1. Define a DbtProject pointing at the dbt project directory (and manifest/state paths) to make dbt models available to Dagster.
2. Decorate a function with @dbt_assets referencing the compiled manifest.json, and have the function accept an AssetExecutionContext and a DbtCliResource parameter.
3. Inside the function, invoke a dbt command via the resource's cli() method and yield from the resulting event stream so Dagster translates each dbt node's success/failure into asset materialization events.
4. Register DbtCliResource in the Definitions object's resources dictionary, pointing project_dir at the dbt project so runs execute the correct dbt invocation.
5. Customize asset keys/groups by subclassing DagsterDbtTranslator and overriding methods like get_asset_key or get_group_name if the default dbt-model-to-asset mapping doesn't match your naming conventions.