Databricks CLI commands

The Databricks CLI speaks to every workspace API surface. In 2026 it is the primary interface for bundle deploys, cluster ops, SQL queries, and file transfers; the web UI is a visualizer.

Install:

brew install databricks      # macOS
# or
pip install databricks-cli   # any platform

Authenticate:

databricks auth login --host https://your-workspace.cloud.databricks.com
databricks current-user me   # confirm

Bundles (most-used surface)

# Scaffold a new bundle
databricks bundle init default-python

# Check everything before deploying
databricks bundle validate [--target dev]

# Deploy to a target
databricks bundle deploy --target dev

# Run a job from the bundle
databricks bundle run <job_name> [--target dev]

# Tear down
databricks bundle destroy --target dev [--dry-run]

# Summarize the bundle
databricks bundle summary --target prod

See the Asset Bundles guide for full coverage.

Workspace + file operations

# List workspace objects
databricks workspace list /Users/your.name

# Export a notebook to your local machine
databricks workspace export /Users/you/notebook.py ./notebook.py

# Import a notebook
databricks workspace import ./notebook.py /Users/you/notebook.py --language PYTHON

# Volume file ops (UC volumes)
databricks fs cp ./init.sh dbfs:/Volumes/prod/shared/init-scripts/install_deps.sh
databricks fs ls dbfs:/Volumes/prod/shared/

Clusters (all-purpose)

# List clusters
databricks clusters list

# Create a cluster from JSON spec
databricks clusters create --json @cluster.json

# Restart
databricks clusters restart <cluster-id>

# Terminate (stops but keeps the config)
databricks clusters delete <cluster-id>

# Permanently delete config
databricks clusters permanent-delete <cluster-id>

# Get event log
databricks clusters events --cluster-id <id>

For job clusters you rarely use these directly; bundle-deployed jobs bring their own cluster config.

SQL warehouses

# List
databricks sql warehouses list

# Inspect one
databricks sql warehouses get <warehouse-id>

# Edit config (JSON patch)
databricks sql warehouses edit <warehouse-id> --json '{
  "size": "MEDIUM",
  "min_num_clusters": 1,
  "max_num_clusters": 5,
  "auto_stop_mins": 5,
  "enable_photon": true,
  "enable_serverless_compute": true
}'

# Start / stop
databricks sql warehouses start <warehouse-id>
databricks sql warehouses stop  <warehouse-id>

Jobs

Most Causeway jobs are bundle-defined; you rarely create jobs by CLI. Useful lookups:

# List jobs visible to you
databricks jobs list

# Get a job's definition
databricks jobs get <job-id>

# Trigger a run
databricks jobs run-now <job-id>

# Recent runs for a job
databricks jobs list-runs --job-id <id>

# Get a specific run's result
databricks jobs get-run <run-id>

Pipelines (Lakeflow Declarative Pipelines)

# List
databricks pipelines list-pipelines

# Get one
databricks pipelines get <pipeline-id>

# Trigger an update (with optional full-refresh)
databricks pipelines start-update <pipeline-id> \
  [--full-refresh] \
  [--full-refresh-selection silver_customers,silver_orders]

# List recent updates
databricks pipelines list-updates <pipeline-id>

# Get one update's details
databricks pipelines get-update <pipeline-id> --update-id <update-id>

Unity Catalog

# List catalogs / schemas / tables
databricks catalogs list
databricks schemas list --catalog-name prod
databricks tables list --catalog-name prod --schema-name silver

# Describe a table
databricks tables get prod.silver.customers

# Grants on a securable
databricks grants get --securable-type TABLE --full-name prod.silver.customers

# Apply a grant via JSON
databricks grants update --securable-type TABLE --full-name prod.silver.customers \
  --json '{"changes":[{"principal":"data-engineering","add":["SELECT"]}]}'

Secrets

# List scopes
databricks secrets list-scopes

# Create a scope
databricks secrets create-scope <scope-name>

# Put a secret
databricks secrets put-secret <scope-name> <key> --string-value "<value>"

# Read
databricks secrets get-secret <scope-name> <key>

Secrets in scopes are the correct home for API tokens, database passwords, and cloud credentials. Reference them from notebooks and jobs with dbutils.secrets.get(scope, key); reference them from bundles with ${secrets.<scope>.<key>}.

Lakebase

# List instances
databricks lakebase list-instances

# Get one
databricks lakebase get-instance <instance-id>

# Create
databricks lakebase create-instance --json @instance.json

# Branch / restore (Autoscaling generation)
databricks lakebase create-branch --source prod-serving --name pr-42
databricks lakebase restore      --at 2026-04-20T03:15:00Z

Auth + config

# List configured profiles
databricks auth profiles

# Change active profile
databricks auth login --host https://... --profile <profile-name>

# Switch context per command
databricks --profile staging clusters list

# Show config file path
databricks auth describe

Profiles live in ~/.databrickscfg. One profile per workspace is the default; workload identity federation (OIDC) replaces the token flow for CI.

Convention: `--json`

Most mutating commands accept a JSON spec rather than a long flag list:

# Via a file
databricks clusters create --json @cluster.json

# Inline (prefer file for anything non-trivial)
databricks sql warehouses edit <id> --json '{"size": "MEDIUM"}'

Keep JSON specs in Git next to the bundle they relate to, not as one-liners in README snippets.

Output formats

# Default: table
databricks jobs list

# JSON (parseable)
databricks jobs list --output JSON

# Specific column
databricks jobs list --output JSON | jq '.jobs[].settings.name'

Debugging

# Verbose logs
databricks --debug jobs list

# Show the HTTP request being made
databricks --log-level DEBUG jobs list

# Show the CLI's config
databricks auth describe