anjijava16/databricks-agents-genai/quickstart

Quickstart & Authentication

Prerequisites

• uv (Python package manager)
• nvm with Node 20 (for frontend)
• Databricks CLI v0.283.0+

Check CLI version:

databricks -v  # Must be v0.283.0 or above
brew upgrade databricks  # If version is too old

Run Quickstart

uv run quickstart

Options:

• --profile NAME: Use specified profile (non-interactive)
• --host URL: Workspace URL for initial setup
• --skip-lakebase: Skip Lakebase setup (non-interactive / CI use)
• --app-name NAME: Existing Databricks app name to bind this bundle to
• -h, --help: Show help

Examples:

# Interactive (prompts for profile selection)
uv run quickstart
 
# Non-interactive with existing profile
uv run quickstart --profile DEFAULT
 
# New workspace setup
uv run quickstart --host https://your-workspace.cloud.databricks.com
 
# Bind to an existing app created via the Databricks UI
uv run quickstart --app-name my-existing-app
 
# Skip Lakebase setup (CI / non-interactive)
uv run quickstart --profile DEFAULT --skip-lakebase

What Quickstart Configures

Creates/updates .env with:

• DATABRICKS_CONFIG_PROFILE - Selected CLI profile
• MLFLOW_TRACKING_URI - Set to databricks://<profile-name> for local auth
• MLFLOW_EXPERIMENT_ID - Auto-created experiment ID

Updates databricks.yml:

• Sets experiment_id in the app's experiment resource
• Updates app name field if --app-name is provided

Existing App

If you created an app via the Databricks UI before cloning a template, use --app-name to bind the bundle to it:

uv run quickstart --app-name my-existing-app

Quickstart will update databricks.yml with the app name and print the binding command:

databricks bundle deployment bind <KEY> my-existing-app --auto-approve
databricks bundle deploy

This avoids the "An app with the same name already exists" error on first deploy.

Idempotency

Re-running quickstart is safe:

• Experiment: If MLFLOW_EXPERIMENT_ID is already in .env and the experiment still exists, it is reused (no duplicate created).
• Lakebase: If Lakebase config is already in .env, the interactive prompt is skipped and the existing config is reused.

Manual Authentication (Fallback)

If quickstart fails:

# Create new profile
databricks auth login --host https://your-workspace.cloud.databricks.com
 
# Verify
databricks auth profiles

Then manually create .env (copy from .env.example):

# Authentication (choose one method)
DATABRICKS_CONFIG_PROFILE=DEFAULT
# DATABRICKS_HOST=https://<your-workspace-here>.databricks.com
# DATABRICKS_TOKEN=dapi....
 
# MLflow configuration
MLFLOW_EXPERIMENT_ID=<your-experiment-id>
MLFLOW_TRACKING_URI="databricks://DEFAULT"
MLFLOW_REGISTRY_URI="databricks-uc"
 
# Frontend proxy settings
CHAT_APP_PORT=3000
CHAT_PROXY_TIMEOUT_SECONDS=300

Next Steps

After quickstart completes:

1. Run uv run discover-tools to find available workspace resources (see discover-tools skill)
2. Run uv run start-app to test locally (see run-locally skill)