Develop DAGsAstro IDE

Astro IDE Quickstart

Private Preview
This feature is in Private Preview.

Get started with Astro IDE by entering a prompt, deploying to an ephemeral test Deployment, and testing your Airflow dag.

Prerequisites

Workspace Author permissions let you create and edit projects with the Astro IDE, but you need Workspace Operator permissions to start an ephemeral test Deployment.

1

Log in and create a project

  1. Log in to the Astro UI and select a Workspace.
  2. Click Astro IDE in the menu. If you are the first person in your Workspace to use the Astro IDE, the Projects page is empty.
  3. On the main Astro IDE screen, you can find options to connect an existing project, Connect GitHub Project or Connect CLI Project. Or, you can enter a prompt in the text box describing what kind of pipeline you want to build. The Astro IDE then generates the project for you.
  4. This quickstart guides you through creating a new project from a prompt.
2

Prompt, review, and accept your first dag

  1. In the Astro IDE, enter the following prompt in the text box describing the Hello World dag you want to build. The more specific the prompt is, the better the agent performs. For this quickstart, enter:
Generate a dag called hello_world_dag that prints 'Hello World' to the logs. The dag should:
- Not require the scheduler to be running.
- Contain a single PythonOperator task named print_hello.

The Astro IDE ephemeral test environment sets AIRFLOW__SCHEDULER__USE_JOB_SCHEDULE=false by default. This means scheduled or downstream tasks do not run automatically. This setting enables the Test tab to give you direct, manual control over when tasks run, instead of relying on Airflow’s scheduler.

  1. The Astro IDE generates the code and display the proposed changes in the suggested changes panel.
  2. Review the changes, and iterate on them with the AI panel. Once you are satisfied with the changes, click the Accept button in the panel to approve the new dag.
3

Deploy and test

  1. After you accept a code change, click Start Test Deployment to test your changes. This deploys the generated code to an ephemeral test Deployment that allows you to run tasks and view logs.

AI can make mistakes, and the generated code can sometimes throw an import error. If you have an import error, the hello_world_dag doesn’t appear in the Test tab, and the IDE lists the error in the Import Errors of the Test Output pane. To fix the error, paste the error message into the prompt box, so the IDE can suggest edits. Review the suggested changes, accept them, and click Sync to Test to update the Deployment.

  1. Open the Test tab and select hello_world_dag from the dag options.
  2. Click Run Task to execute the dag.
  3. Confirm the task ran successfully by checking the Task Logs.

You can also view your ephemeral test Deployment in the Airflow UI by clicking Open Airflow from the toolbar menu.