---
title: 'Airflow, now available within ChatGPT'
description: >-
  Astronomer has just released the astronomer-data plugin for ChatGPT, which
  makes it possible to manage Airflow from within ChatGPT Work or Codex.
date: 2026-10-01T19:38:55.573Z
authors:
  - author: src/content/people/andrew-doberstein.md
canonical_url: 'https://www.astronomer.io/blog/airflow-now-available-within-chatgpt/'
---
In collaboration with OpenAI, Astronomer is announcing the new astronomer-data plugin for ChatGPT. ChatGPT users who install this plugin will now be able to interact with their Airflow environment without leaving Codex or ChatGPT Work, even if they’re running self-hosted OSS Airflow.

By making it easier to manage Airflow from within a workflow application you already use, we’re hoping to reduce the burden of switching tools and ultimately make your time spent managing Airflow even more productive. Now you can stay in the flow while still effectively managing your data pipelines to ensure your data is fresh, accurate, and useful.

> “Data pipelines are at the heart of most modern businesses, and Airflow is one of the leading open-source standards for managing them. Astronomer’s plug-in for Airflow lets data engineers use natural language within ChatGPT to design, troubleshoot, and upgrade Airflow so that critical data is reliably delivered to every part of the business that needs it.” **— Arpan Shah, General Manager, Technology at OpenAI**

## How it works in action

**Jordan** is a data engineer. The `orders_daily` Dag failed overnight, and it's the first thing waiting when they open ChatGPT.

*“Why is my orders\_daily Dag failing? Find the root cause and fix it.”*

The plugin pulls the actual task logs from the Airflow deployment, traces the failure to a schema change in an upstream table, and proposes a fix to the task code. Jordan reviews the diff, approves it, and redeploys, all without opening a terminal.

Later that day, Jordan needs a new pipeline entirely.

*"Write a Dag that loads new S3 files into Snowflake every hour."*

The plugin scaffolds the Dag using Jordan's team's existing conventions, and Jordan asks a follow-up to check its work:

*“Did it run correctly? Show me row counts from the last three runs.”*

The plugin queries the warehouse directly and confirms the pipeline is working, in the same conversation, with no dashboard to build or separate SQL client to open.

## What the plugin can do

* **Build and test Dags.** Write new pipelines, extend existing ones, and validate them locally before deploying, including dbt projects run through Cosmos.
* **Debug failures.** Read task logs, find the root cause of a Dag that won't parse or a task that keeps failing, and fix the underlying code.
* **Run and manage Astro.** Set up a new Astro project, troubleshoot a deployment, or check on system health.
* **Migrate.** Move Dags from Airflow 2 to Airflow 3, or bring a project over from another orchestrator.
* **Query the warehouse.** Check freshness, profile a table, or confirm a pipeline actually did what it was supposed to, in Snowflake, BigQuery, Postgres, or 25+ other connected databases.
* **Trace lineage.** See what data feeds a table, and what breaks downstream if it changes.

## Founded on open source

This plugin is made possible because of the [Airflow MCP server](/docs/astro/astro-mcp-server) and the library of skills published in the [astronomer/agents repository](https://github.com/astronomer/agents), which is available to Astro customers and open-source Airflow users alike. We remain committed to contributing to the Airflow open source community. That said, there are still a lot of great reasons to choose Astronomer as a managed solution. With us, you get access to Otto, our specialized data engineering agent, as well as stress-free deployments, world-class support, and observability across all of your Airflow deployments.

## Get started today

Install the astronomer-data plugin in the ChatGPT marketplace to connect it to your Airflow deployment and your warehouse. While the plugin functions with both ChatGPT Work and Codex, we recommend using Codex because many of the skills run local tools such as the Astro CLI. Getting started is as easy as following the link below!

[Install the astronomer-data plugin →](https://chatgpt.com/plugins/plugins_6ab2f25e4928819184294ebadcbe38ab)