The Astro CLI is the fastest way to develop and deploy Airflow pipelines from your own editor. Run astro dev start and you get a local Airflow environment with hot-reloading, an Airflow UI that opens automatically, and Dag changes that reflect immediately without manual restarts. When you're ready to ship, astro deploy builds your project image, authenticates to Astro, and pushes to your deployment in a single command.

Today we're making it more powerful. AI coding agents are only as capable as the tools they can reach. If an agent can't access your APIs, can't spin up a local environment quickly, or has to work around tooling built for humans, you end up doing the translation work yourself.

That's what this release is about. We're shipping a set of additive, non-breaking improvements that make the Astro CLI faster for developers and more accessible for agents. Here's what's new.

Faster local development: standalone mode and automatic port management

Two improvements in this release reduce friction for local Airflow development.

Standalone mode gives developers a lightweight option for running Airflow locally without a container runtime. This means you can get up and running with Airflow in seconds and not minutes. It leverages the airflow standalone mode and uses uv-managed Python environments and SQLite, which makes it a good fit for anyone developing outside the context of an Astro project or working in environments where Docker or Podman isn't available. To use it, run astro config set dev.mode standalone and then use astro dev as you normally would.

Built-in reverse proxy removes port collision headaches when running multiple Astro projects at the same time. Every project gets a .localhost hostname routed automatically. Port 8080 stays the default for your first project, and subsequent projects get ports allocated in the background. You access everything through clean URLs:

# Before: manually set ports per project to avoid collisions
$ astro config set webserver.port 8081

# After: just run astro dev start in each project
➜ Airflow UI: http://alpha.localhost:6563
➜ Airflow UI: http://beta.localhost:6563

No config changes. No port juggling. Single-instance users see no difference.

Direct API access from the CLI: astro api

The new astro api command gives you structured access to both the Astro API and the Airflow REST API directly from the CLI.

# List Astro platform endpoints
astro api cloud ls

# Call a platform endpoint
astro api cloud list-deployments --path-param organizationId=<org-id>

# List Airflow endpoints on a deployed instance
astro api airflow -d <deployment-id> ls

# Call an Airflow endpoint
astro api airflow -d <deployment-id> get_dags

Two subcommands:

• astro api cloud — Authenticated requests to the Astro platform API using your current context's bearer token.
• astro api airflow — Requests to the Airflow REST API (local or deployed), with automatic version detection and OpenAPI spec resolution.

Both support endpoint discovery via ls and describe, pagination, JSON output for piping to tools like jq, Go-template output formatting, magic field syntax for request bodies (@file, :=json, =string), and a --curl flag that emits the equivalent curl command for any request.

Why this matters: AI coding agents (Cursor, GitHub Copilot, Claude Code, and others) interact with your stack through the CLI. Until now there was no direct, structured path for an agent to reach the Airflow or Astro APIs. astro api closes that gap. An agent can discover endpoints, call them, and handle responses without a human translating between tools.

Where we're headed

These updates reflect a deliberate direction: we're investing in a CLI that works better for data engineers and for the AI agents working alongside them. Faster local environments, direct API access, and structured outputs benefit both. As AI agents become a bigger part of data engineering workflows, we want Astro to be the platform that supports how modern teams actually work, whether that's a developer iterating locally or an agent running a task autonomously.

This is a foundation, and there's more investment coming. If you want to see what's possible today, check out the Astro CLI documentation here.