Skip to main content

Ship an app from your AI editor

NovaNova
Deploy <registry>/latency-api:<tag> to prod-01.
Ask Nova

Rather not set up the editor? If you already have an image, ask Nova this. It shows the plan and deploys it when you approve.

Bob is writing latency-api in an AI editor and wants it running on prod-01 today, without stopping to learn Kubernetes. So Bob types "deploy this", reads the plan that comes back and approves it, and nobody writes a Dockerfile. Alice on the platform team grants deploy access once and still holds the controls.

Before you start​

  • A cluster, prod-01 here, created or registered.
  • An MCP editor (Claude Code, Cursor or VS Code) open on your project, latency-api here.
  • astroctl, the AstroPulse CLI, signed in on Bob's machine; the deploy server uses it.

1. Alice requests deploy access​

Path: AI Agent > Nova AI > Nova Connect

  1. On the Prompt to Production card, select Request Access.
Once per organization

Once granted, every developer in the organization sees Download astro-deploy here.

2. Bob connects the editor to AstroPulse​

Path: AI Agent > Nova AI > Nova Connect > Integrations

  1. Pick your editor and add the endpoint it shows, https://platform.astropulse.io/mcp.
  2. Sign in when the editor opens the browser.

You should see: AstroPulse's tools listed in the editor.

3. Bob adds the deploy server​

Path: AI Agent > Nova AI > Nova Connect > Prompt to Production

  1. Select Download astro-deploy and install it, such as pipx install ~/Downloads/astro_deploy-*.whl.
  2. In the latency-api project, add it to the editor:
Terminal
claude mcp add astro-deploy -s project -- mcp-server-astro-deploy
Per project

-s project adds the deploy server to this project only. Run it in each project you ship from.

4. Bob asks for the plan​

Path: Your editor > latency-api

  1. Tell the editor: Deploy this to prod-01.
  2. Read the plan.
The plan is the gate

Nothing exists on prod-01 until Bob says go.

5. Bob approves and sees it running​

Path: Workloads > Applications > latency-api

  1. Tell the editor to go ahead.
  2. Open latency-api in the console.

You should see: latency-api Running, with its Events, Logs and History.

NovaNova
Is latency-api running on prod-01?
Ask Nova

It reads the app's status and health.

Clean up​

On the latency-api page select Delete, type the name, then Delete Permanently.

NovaNova
What does deleting latency-api remove?
Ask Nova

It explains what goes and gives you the command to run. Nova never deletes an app itself.

Next: Deploy an open model