The Dagster CLIs

We built CLI tooling before AI-driven workflows really took off. It was our attempt to assist with scaffolding, building, and designing Dagster projects with a CLI that encodes opinionated choices about project structure and patterns so you get the most out of Dagster.

create-dagster CLI (bootstrapping)

The first step in building with Dagster is building a project. The create-dagster CLI is the official way to scaffold that initial structure:

  • create-dagster project — Scaffold a new Dagster project at a given path. Creates src/PROJECT_NAME/ with definitions.py, a defs/ directory.
  • create-dagster workspace — Initialize a new Dagster workspace with a projects/ folder, deployments/local/, and a root dg.toml. Use this when you want to manage multiple projects in one repo.

Run from an empty directory or pass . to create in the current directory. For installation and full options, see the create-dagster CLI reference.

dg CLI (developing)

The dg CLI gives you deterministic actions over your Dagster project. Instead of having to guess where to put files or how to validate definitions, dg offers consistent, correct behavior.

There are a number of dg commands:

  • dg scaffold: scaffold Dagster entities (e.g. assets, components), GitHub Actions workflows, and build artifacts.
  • dg dev: start a local Dagster instance.
  • dg check: check the integrity of your Dagster code.
  • dg list: list Dagster entities in the current environment.
  • dg launch: launch a Dagster run (e.g. by asset selection, job, or partition).
  • dg api: make REST-like API calls to Dagster Plus (agents, assets, deployments, runs, logs, schedules, sensors, secrets, and more).
  • dg plus: commands for interacting with Dagster Plus—login, deploy (build and push, configure, start/finish), create (e.g. CI API tokens, env vars), and pull (e.g. env from cloud).

Dagster Components (integrate)

Dagster Components work hand-in-hand with dg to let you build pipelines in a lightweight way. Components provide an easy interface for designing Dagster applications as well as prebuilt solutions for many common integrations.

Why it matters for AI-driven workflows

Instead of writing Dagster code directly, using dg and Components allows for a more consistent coding experience and makes it easier to do common things like:

  • Scaffold new assets or other definitions in the correct location
  • Validate that definitions load and pass checks
  • List what's defined and launch runs

That keeps behavior predictable and aligned with how we recommend structuring Dagster projects. We'll see in later sections how the Dagster Skills use dg as the primary way to interact with your project.