Install a Release
Install Envy with Helm, then open the dashboard. You do not need the Envy CLI,
Go, or a source checkout. The same chart works on Docker Desktop Kubernetes
and remote clusters, and pulls the prebuilt davey/envy:0.3.0 image.
Before you start
Section titled “Before you start”- A running Kubernetes cluster, locally or remotely, and permission to install workloads and RBAC.
-
kubectlconfigured for that cluster and Helm (brew install helmon macOS). - PostgreSQL reachable from the cluster, with its connection URL ready.
- Your chosen Istio, Cilium, or Linkerd mesh installed; see Choose Your Mesh.
- Network access to Docker Hub to download the released chart and images.
The steps below create the database Secret and Helm values. For the first look, you can use a local port-forward and password login. Public DNS/TLS, Google sign-in, and an application baseline can follow before sharing the installation and creating real previews.
Install with Helm
Section titled “Install with Helm”With your cluster settings saved in values.yaml, the installation command is:
helm upgrade --install envy oci://registry-1.docker.io/davey/envy-chart \ --version 0.3.0 --namespace envy-system --create-namespace \ --values values.yaml --wait --timeout 5mThe steps below prepare that file and open the UI. If you only want to see what Envy looks like, start with the dashboard walkthrough.
1. Prepare your cluster
Section titled “1. Prepare your cluster”You need Helm, kubectl access to a Kubernetes cluster, and
PostgreSQL reachable from that cluster. For Docker Desktop, enable Kubernetes
and select its context; Docker alone is not a Kubernetes cluster.
kubectl config current-context# For Docker Desktop, if that is your intended cluster:# kubectl config use-context docker-desktopEnvy uses your existing Istio, Cilium, or Linkerd mesh for previews. Follow Choose Your Mesh for your cluster’s profile. The chart installs Envy’s server, bundled web UI, database migrations, Service, and permissions. It does not install PostgreSQL or the mesh.
Create the namespace and database Secret before running Helm. Save your actual
PostgreSQL connection URL in a private file named database-url, then run:
kubectl create namespace envy-systemkubectl -n envy-system create secret generic envy-database \ --from-file=url=./database-urlSkip creation if the namespace or Secret already exists. Use the connection URL and TLS settings supplied by your database operator. Keep this file out of Git.
2. Save your settings and install
Section titled “2. Save your settings and install”Save this as values.yaml for a first look through a local port-forward on an
Istio cluster:
installationID: envy-evaluationexternalDatabase: secretName: envy-database secretKey: urlmesh: provider: istioauth: mode: password externalOrigin: http://localhost:8081For Cilium or Linkerd, use the matching mesh settings instead, retaining the evaluation authentication settings above. Choose the mesh and a stable installation ID before the first install; they are recorded in the database and cannot be switched on a populated installation.
values.yaml is Helm’s configuration file. There is no installation.json
requirement and no CLI command to run first. That separate file is only used
by the optional installation diagnostic.
Now run the Helm command at the top of this page. It downloads the versioned chart and starts the released image; there is nothing to build locally.
3. Open the dashboard
Section titled “3. Open the dashboard”Once Helm finishes, forward the Envy Service to your laptop:
kubectl -n envy-system port-forward service/envy-envy 8081:8081Keep that command running and open http://localhost:8081.
This works for a remote cluster too. The UI and API share this address.
Use localhost consistently: the configured authentication origin must match
the browser address, including the port.
With the evaluation settings above, sign in as admin with password admin.
Set your own password or Google sign-in before sharing
the installation. No Vite server or CLI login is needed to use this bundled UI.
Your new installation starts with an empty catalog. You can look around, or select Explore demo on the login screen to view sample previews. Demo simulation does not deploy workloads or create working preview endpoints.
4. Create your first real preview
Section titled “4. Create your first real preview”When you are ready to connect an application:
- Complete your mesh and preview ingress settings, including preview DNS/TLS, and rerun Helm with the updated values.
- Configure authentication and a public HTTPS origin if other people will use the installation.
- Onboard a running application, then create and inspect a preview in the web interface.
GitHub automation, agent connections, and CLI installation can follow when you need them.
Upgrades and troubleshooting
Section titled “Upgrades and troubleshooting”Pin the chart version and retain your values file. To upgrade, review the target release notes, back up PostgreSQL, and rerun the same Helm command with the new version. The chart selects its matching image automatically.
If Helm times out, check pods and migration Job logs in envy-system; a missing
database Secret or unreachable database prevents installation. See the
installation reference
for detailed networking and Secret settings. Helm rollback does not reverse
database migrations; see the operations guide.