Skip to main content
Ask your coding agent to deploy a project from GitHub. If your Qovery organization has no cluster yet, the agent creates one on your laptop with Docker, builds the project and deploys it, and gives you a URL. You need no cloud account, no credit card and no Kubernetes knowledge. When you are ready for production, you create a cluster in your own cloud account and deploy the same project to it.
The local cluster runs on your machine only. It is for trying Qovery and testing, not for production.

​
What you need

  • Docker, running, with at least 4 CPUs and 8 GB of memory.
  • macOS, Linux, or Windows with WSL.
  • k3d, helm and kubectl. On macOS: brew install k3d helm kubectl. The agent checks for them and tells you what is missing.
  • The Qovery CLI, version 1.180.1 or later. See Install the CLI.
  • A Qovery account. The Free plan is enough, with no credit card. Sign up at console.qovery.com. The local cluster counts as a cluster on your plan, and the Free plan includes one.
  • A coding agent that supports Agent Skills: Claude Code, Cursor, OpenCode, Gemini CLI and others.

​
1. Install the Qovery skills

The skills teach your agent the Qovery CLI and API, including how to create the local cluster. See Agent Skills for what each one does.

​
2. Ask your agent

Open your agent and paste:
claude
traefik/whoami is a small public web server, good for a first run. Replace the URL with any repository that has a Dockerfile, or with the folder you are working in.

​
3. Answer the agent’s questions

The agent does the work, but a few steps need you, because they open a browser or ask for your password: In Claude Code, type ! before a command to run it in the session, for example ! qovery auth. Then the agent:
  1. Checks Docker, k3d, helm, kubectl and the CLI version.
  2. Runs qovery demo up, which creates a k3s cluster with k3d, installs the Qovery Operator on it and connects it to your organization.
  3. Creates a project and an environment, adds the repository as an application, builds it on the cluster and deploys it.
  4. Gives you the application URL.

​
How long it takes

One run on an Apple M3 Ultra with Claude Code, in November 2026: Your times depend on your machine, your network and the project you deploy.

​
What to try next

Your own project

Open your agent in your repository and say “Deploy my application with Qovery”. The agent writes a Dockerfile if the project has none.

An open-source project

Deploy any open-source project that ships a Dockerfile, the same way as traefik/whoami. Give the agent the repository URL and the environment variables it needs.

Watch it in the Console

Your local cluster, the environment and the deployment logs show up in the Console like any other cluster.

Prefer to do it by hand?

Test & Learn Locally walks through the same setup with the CLI, without an agent.

​
Go to production

The local cluster is for testing. For production, create a cluster in your own cloud account and deploy the same project to it. On the Free plan, which includes one cluster, remove the local cluster with --delete-qovery-config first, or move to a paid plan to keep both. Ask your agent “Deploy this to production on AWS with Qovery”, or follow a guide:

Create a cluster in your cloud

AWS, GCP, Azure or Scaleway. Qovery creates and manages the Kubernetes cluster in your account.

Bring your own cluster

Connect a Kubernetes cluster you already run: EKS, GKE, AKS, Kapsule or on-premise.

​
Remove the local cluster

Ask your agent to remove it, or run:
This deletes the local cluster and keeps its record in Qovery. To also delete the record and the environments deployed on it, run qovery demo destroy --delete-qovery-config. If the cluster was created with --cluster-name, pass the same name.

​
If something goes wrong

qovery demo up ends with an error code, such as DOCKER_NOT_RUNNING or MISSING_DEPENDENCY, and the agent knows what to do with each one. If it still fails, see Local Cluster for Development or ask in the Qovery community.