(feat): Add Genie Space resource support for Databricks Asset Bundles [DIRECT mode only]

[nijanthanvijayakumar]

Changes

Adds genie_spaces as a new bundle resource type for deploying AI-powered Genie Spaces via DABs but supporting only direct mode (and not Terraform deploy).

Resource fields: title, description, parent_path, warehouse_id, serialized_space

Features:

• Direct deploy CRUD operations (create, update, delete).
• Dev mode prefix (auto-prefixes title with [dev username]).
• Default parent path to workspace.resource_path.
• Bundle-level permissions: CAN_MANAGE, CAN_VIEW, CAN_RUN (mapped from bundle's permissions: block).
• Resource-level permissions: CAN_MANAGE, CAN_EDIT, CAN_RUN, CAN_VIEW (set directly on resource).

Example:

resources:
  genie_spaces:
    sales_assistant:
      title: "Sales Assistant"
      description: "Ask questions about sales data"
      warehouse_id: abc123def456
      serialized_space: '{"version":1,"config":{"sample_questions":[{"id":"a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4","question":["Show top 10 sales"]}]},"data_sources":{"tables":[{"identifier":"catalog.schema.sales"}]}}'
      permissions:
        - level: CAN_VIEW
          group_name: data-analysts

Requires direct deploy: DATABRICKS_BUNDLE_ENGINE=direct databricks bundle deploy

Notes: CAN_EDIT has no bundle-level equivalent, so it seems that it has to be set directly on the resource.

Why

Solves issue: #3008

Genie Spaces are AI-powered data rooms that let users ask natural language questions about their data. Users requested the ability to manage Genie Spaces as code via DABs for version control and CI/CD workflows.

Direct deploy only: Terraform mode is not supported because the Databricks Terraform provider does not yet have a databricks_genie_space resource. Once the provider adds support, Terraform mode can be enabled.

Tests

• Unit tests for resource type, permissions, and state management.
• Acceptance tests for deploy/destroy lifecycle with mock server.
• Manual testing on Databricks workspace (screenshots below).

A view of the databricks.yml.

Bundles deployment and validation to Databricks Free Edition

Genie Space creation success ✅

A view of the databricks.yml for testing the update to Genie Space

Bundles deployment and validation to Databricks Free Edition (second try)

Genie Space update success ✅

Bundle destroy

[nijanthanvijayakumar]

Hi @andrewnester - I am not sure what a cla-required label is - is that a Contributor Licensing Agreement and is there something that I have to do from my end?

[andrewnester]

Hi @nijanthanvijayakumar ! Thanks for the PR! Indeed, contributing to Databricks CLI requires signing CLA, so if that's something you're up for doing, please write a message to [email protected] with CLA sign request and we'll take it from there

[nijanthanvijayakumar]

Hi @nijanthanvijayakumar ! Thanks for the PR! Indeed, contributing to Databricks CLI requires signing CLA, so if that's something you're up for doing, please write a message to [email protected] with CLA sign request and we'll take it from there

Yeah of course @andrewnester - if there is a template/format that I could follow to send an email, I'd be happy to. Thanks for the reply once again!

[andrewnester]

@nijanthanvijayakumar, something like this is just enough

Subject: CLA signing

Hey Team,
I would like to sign a CLA in order to contribute to Databricks CLI.

[github-actions]

An authorized user can trigger integration tests manually by following the instructions below:

Trigger:
go/deco-tests-run/cli

Inputs:

• PR number: 4191
• Commit SHA: eabf9d93085717535ce596e0eb77ccb4176e8eb0

Checks will be approved automatically on success.