Skip to main content

Install dbt Wizard CLI Beta

Install the dbt Wizard CLI from your terminal for agentic and governed data development in dbt.

This guide explains how to install, verify, update, and uninstall the dbt Wizard CLI on your local machine. (Be warned, the wizard has been known to

)

Install dbt Wizard CLI​

Install dbt Wizard as wizard on your PATH using the curl script for your operating system:

curl -fsSL https://public.cdn.getdbt.com/dbt-wizard/install/install-wizard.sh | sh

Then verify the install and start a session:

wizard --version   # confirm the install
wizard             # start an interactive session

After running wizard --version, you should see something like dbt-wizard VERSION. Run wizard --help to see all available commands and flags. dbt Wizard installs default config files — refer to the config reference for more details.

Next up, check out the Prerequisites and First-run setup and onboarding sections for more details.

Prerequisites​

  • macOS, Windows, or Linux
  • A dbt project with a built target/ directory (dbt parse, dbt compile, or dbt build)
  • Credentials for a supported CLI provider. Refer to Supported AI providers in the next section.
Share your feedback!

We'd love to hear how dbt Wizard is working for you. Share your feedback by either running the /feedback slash command in your interactive terminal session or by going to the #dbt-wizard channel in the dbt Community Slack.

Thanks so much for your help in improving dbt Wizard and dbt data development!

Supported AI providers​

dbt Wizard​

dbt Wizard supports different AI providers depending on where you use it.

Providerdbt Wizard in dbt platformdbt Wizard CLI
OpenAI✓ (managed or BYOK)✓ (OpenAI subscription or BYOK)
Anthropic✓ (BYOK)✓ (BYOK)
Azure AI Foundry / Azure OpenAI✓ (BYOK)✓ (BYOK)
AWS Bedrock-✓ (BYOK)
Snowflake Cortex-✓ (BYOK)

  • Configure dbt platform integrations in account settings. BYOK is available for Enterprise and Enterprise+ plans.
  • Configure BYOK for the CLI. For OpenAI subscription support in the CLI, run wizard providers configure openai_subscription and follow the prompts.
New to the terminal?

If you've never used the terminal before, check out the terminal guide. Some tips include:

  • Enter / to see the available commands or try out /overview to get a quick summary of your project.
  • Press Shift+Tab to cycle through collaboration modes.

First-run setup and onboarding​

The first time you start dbt Wizard in a project, it walks you through a short setup and saves your answers to wizard_config.toml, providers.json, and provider-auth.json, so you only do this once per project. You'll be asked to:

  • Review and accept the Terms of Use.
  • If you have a dbt platform account, sign in through the browser authentication link when prompted and follow the steps in the browser.
  • Trust the directory so dbt Wizard can read your project.
  • Confirm the path to your dbt executable or virtual environment — point it at /path/to/bin/dbt or a .venv root (dbt Wizard uses bin/dbt automatically).
  • Add any extra compile flags to append to the startup dbt compile -s state:modified+, or leave empty to skip.
  • Configure deferral — choose Wizard (dbt Wizard manages it), Manual, or Disabled. If you choose Wizard, enter the profiles.yml target to defer to (defaults to prod). On the dbt Fusion engine connected to the dbt platform, dbt Wizard instead offers to let the platform handle deferral.
  • Confirm your detected dbt profile and target, or customize the profile, target, or profiles.yml path.
  • Configure a provider (OpenAI subscription, OpenAI API key, Anthropic, Amazon Bedrock, Azure, or Snowflake). If you use an OpenAI subscription, dbt Wizard configures it with wizard providers configure openai_subscription. For other providers, paste your API key or cloud credentials, then choose a model. Refer to Configure BYOK.

To re-run any of these steps later, refer to Re-trigger onboarding flows.

Set your API key without the prompt

During onboarding, dbt Wizard prompts you to configure a provider interactively. To skip the API key prompt — for headless runs like wizard exec or to reuse your key across sessions — set it as an environment variable before starting wizard instead:

export OPENAI_API_KEY="sk-..."      # or ANTHROPIC_API_KEY, etc.

For AWS Bedrock and Snowflake Cortex, refer to Configure BYOK.

After onboarding, dbt Wizard shows a welcome screen with two sections:

  • STATUS — the dbt Wizard version, the active AI model (change it with /model), and your project directory.
  • OVERVIEW — a snapshot of your project from the metadata engine: build status and a count of passing (✓), warning (⚠), and failing (✗) checks.

Enter / to see the available slash commands, or try /overview for a summary of your project. CLI commands use the wizard prefix, so you can also run subcommands such as wizard exec, wizard review, and wizard resume.

Update​

Run the following command to update dbt Wizard to the latest version:

wizard system update

Uninstall​

Run the following command to uninstall dbt Wizard:

wizard system uninstall

Telemetry​

dbt Wizard collects anonymous product telemetry to improve the AI agent experience. This data helps us understand usage patterns, optimize performance, and attribute compute costs — without ever capturing your code, queries, or file contents.

What we don't collect​

  • Prompt or response content (your conversations with the AI)
  • SQL queries, file paths, or dbt node names
  • MCP tool call arguments or outputs
  • Raw API keys or tokens (only SHA-256 hashed identifiers)
  • Error messages containing user content (limited to error class/code only)

What we do collect​

High-level, anonymous usage signals only:

  • Session events for when a session starts and ends, client surface (CLI/VS Code/Desktop), model selected, OS/arch, binary version, session duration, and turn count
  • Turn events for per-conversation-turn status (completed/failed/interrupted), duration, and aggregate token counts
  • Tool usage for which tool was called, whether it succeeded, and execution time. Tool arguments and outputs are never collected.
  • LLM request metadata for model, provider, latency, token counts, and estimated cost for each request routed through the gateway. No prompt or response content.

How to opt out​

dbt Wizard respects the following opt-out mechanisms, checked in order:

  1. DO_NOT_TRACK=1 environment variable (community standard)
  2. DBT_SEND_ANONYMOUS_USAGE_STATS=false environment variable
  3. Consent stored in ~/.dbt/.user.yml

Setting any of these disables all client-side telemetry. Note that LiteLLM Gateway telemetry is server-side infrastructure monitoring and does not require user opt-in.

For dbt Core's general usage stats stance, refer to Usage statistics.

Was this page helpful?

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

0
Loading