← Documentation

Quick start

This guide gets you from zero to a connected cluster in Kaja. You add the cluster in Kaja first, then run one command on your infrastructure so you can develop and deploy from the dashboard. You bring your own infra—we don't sell you clusters.

What you need

  • A Kaja account (sign up at the app).
  • A cluster—or a way to create one. For example, k3d runs a small cluster on your laptop in a few seconds.
  • kubectl and Helm 3 installed so you can run the install command.

Step 1: Add the cluster in Kaja

  1. Log in to the Kaja app (e.g. app.kaja.dev).
  2. Choose your organization (or create one—your team's workspace).
  3. Open Clusters and click Connect Cluster.
  4. Enter a name (e.g. "My dev cluster")—a friendly label. Kaja generates a unique resource id for the cluster, same as elsewhere in the app.
  5. Click Connect.

Kaja shows you a setup page with a custom install command. That command contains a one-time token—the agent uses it once to register; after that it's invalid and can't be reused, so there's no long-lived secret to protect.

Step 2: Create your cluster (if you don't have one)

If you're using k3d, create a cluster, for example:

k3d cluster create mydev

If you already have a cluster, skip to the next step.

Step 3: Run the install command on your cluster

On the Kaja setup page you'll see one or two ways to install (e.g. "From a release tarball" and "Or from OCI"). Pick one, copy the whole command, and run it in your terminal. Make sure kubectl is pointing at the cluster you want to connect.

The command installs a small agent on your cluster. The agent talks to Kaja so you can develop and deploy from the dashboard—environments, apps, logs, metrics—on the infra you brought.

When the agent is running, Kaja shows the cluster as Active. You're done.

Step 4: Do something with it

Once the cluster is connected, try:

  • Environments — Create an environment (e.g. "staging" or "dev"). Environments are where you'll put your apps.
  • Container apps — Deploy an app into an environment. Pause it when you're not using it to save cost.

Next steps