Deploy from GitHub

Is your code on GitHub? Connect the repo to your project. Cubli builds it straight away, and then on every push to your branch. That works with any AI tool that puts code in GitHub. You don’t need a Dockerfile or a workflow: see How Cubli builds your code.

Connect a repo

1. Open your project. In an empty project, choose GitHub repo. Already something in it? Right-click the canvas in the advanced view and choose GitHub repo.
2. The first time, click Connect GitHub. GitHub opens in a new tab. Choose which repos Cubli may see there, and come back. You do this once per workspace.
3. Search your repo. Missing one? Click Give Cubli access in GitHub and add it there.
4. Choose the Branch, for example main, and a Name for the app.
5. Is your app in a subfolder, as in a monorepo? Fill in Folder (optional). Empty means the whole repo.
6. Leave Deploy on every push to main on, and click Place on canvas and deploy.

Cubli builds the latest commit and makes it live on a Cubli address. The build log is in the app’s Deploys tab.

On every push

When you push to the chosen branch, Cubli builds that commit and makes it live. The commit message shows with the version. If building or starting fails, the previous version stays online.

Change the connection

Open the app, choose the Settings tab and go to Source:

• Deploy latest commit: builds your branch’s latest commit now, without a push.
• Branch and Folder: apply from the next deploy.
• Deploy on every push: switch it off and you only deploy with the button.
• Stop building from GitHub: click Disconnect. The live version stays. New versions then come through the CLI or your AI tool.

Go back to an earlier version

In the Deploys tab, click Make this version live next to an earlier version. Nothing is rebuilt. Note: the next push goes live as usual. To really remove a mistake, revert the change in GitHub too.

Staging, promoting and rolling back are covered in Staging, promote and roll back.

No GitHub?

No need. With the CLI or your AI tool you put a folder on your computer online. See Your first deploy with the CLI.