There’s partial guidance in the knowledge, but nothing specific to Databricks Asset Bundles, so I’ll show you the closest working pattern: running

edr report

from a Databricks notebook using

%sh

and a dbt profile, which you can adapt into your bundle job/task. ### 1. Ensure packages are installed on the cluster In a notebook cell:

python %pip install elementary-data %pip install "elementary-data[databricks]"

This is required so the

edr

CLI can talk to Databricks. [[Databricks install](https://elementary-community.slack.com/archives/C02CTC89LAX/p1680765973633019?thread_ts=1680765973.633019&cid=C02CTC89LAX)] ### 2. Make sure you have an

elementary

profile You need a

profiles.yml

with an

elementary

profile that points to the schema where Elementary dbt models live. You can generate the profile from your dbt project with:

bash dbt run-operation elementary.generate_elementary_cli_profile

Then copy the output into

profiles.yml

(e.g. in your repo or workspace) and fill in secrets like token, host, http_path. [[Generate profile](https://docs.elementary-data.com/oss/guides/generate-report-ui)] In Databricks, you’ll typically: - Store

profiles.yml

in the repo/workspace path, and - Point

edr

to it with

--profiles-dir

. ### 3. Run

edr report

from the notebook Here is a minimal notebook cell that runs the report against a dbt project checked out in the workspace (paths are examples; adjust to your bundle’s repo path):

python # Python cell that shells out to edr dbt_project_dir = "/Workspace/Repos/<user>/<repo>/path_to_dbt_project" profiles_dir = dbt_project_dir  # or another folder where profiles.yml lives target_dir = f"{dbt_project_dir}/edr_target"  cmd = f""" cd {dbt_project_dir} edr report \   --project-dir "{dbt_project_dir}" \   --profiles-dir "{profiles_dir}" \   --target-path "{target_dir}" """  print(cmd)

python # Actually execute the command result = dbutils.notebook.run("dummy", 0)  # placeholder to keep cell type Python

Or more simply with `%sh`:

bash %sh cd /Workspace/Repos/<user>/<repo>/path_to_dbt_project  edr report \   --project-dir "/Workspace/Repos/<user>/<repo>/path_to_dbt_project" \   --profiles-dir "/Workspace/Repos/<user>/<repo>/path_to_dbt_project" \   --target-path "/Workspace/Repos/<user>/<repo>/path_to_dbt_project/edr_target"

This pattern (using

--project-dir

,

--profiles-dir

,

--target-path

) is what users successfully run from Databricks notebooks. [[Shared-cluster example](https://github.com/elementary-data/elementary/issues/1584)] The HTML report will be created under

edr_target/elementary_report.html

by default (or under the folder you pass via

--target-path

). [[Report location](https://elementary-community.slack.com/archives/C02CTC89LAX/p1687983933201809?thread_ts=1687983933.201809&cid=C02CTC89LAX)] ### 4. (Optional) Upload to S3 or other storage One common pattern is to generate the report in the notebook and then upload it to S3 (or another static hosting location) using Python (e.g.

boto3

) or cloud CLIs, instead of

edr send-report

. [[S3 upload pattern](https://elementary-community.slack.com/archives/C02CTC89LAX/p1681955298320329?thread_ts=1681955298.320329&cid=C02CTC89LAX)] --- To integrate with Databricks Asset Bundles, wire this notebook as a task in your bundle’s job; the core logic is the same: install

elementary-data[databricks]

, ensure

profiles.yml

is available, and call

edr report

with the correct paths. The knowledge sources don’t provide a bundle-specific YAML example, so I can’t give a fully accurate bundle config snippet.