Getting started

Connect a cluster without guessing.

KrevoPilot generates a cluster-specific Helm command. Follow this sequence so the saved connection, key and incoming data stay attached to the same cluster history.

Typical setup: about 3 minutesHelm-basedBash and PowerShell

Sign in and confirm access

Open app.krevopilot.com. An Owner or Admin can create cluster connections. A team member also needs access to the relevant cluster before it appears in their dashboard.

New customer?

Use the invitation from your workspace owner, set your password, then sign in. SSO users can use the company sign-in button after the workspace has enabled OIDC.

Check prerequisites

kubectl accessYour context points to the intended cluster.

Helm 3Installed on the machine running the command.

Outbound HTTPSThe cluster can reach https://api.krevopilot.com.

Install permissionYou may create the agent namespace and RBAC resources.

kubectl config current-context
helm version

Confirm the context name before continuing. Installing into the wrong context connects the wrong cluster.

Create the saved cluster connection

Open Clusters and select Connect cluster.
Enter a recognizable cluster name and stable alias.
Continue to installation. KrevoPilot creates a unique one-time key for this saved cluster record.
Select Bash or PowerShell. Copy the complete generated command.

Protect the key

It authenticates the agent. Do not paste it into tickets, screenshots, chat, source control or shell history shared with others. Generate a new key if it is exposed.

Install the generated command

Run the exact command shown by the app in the cluster you selected. The command contains the current chart, image, platform URL, cluster alias and one-time key. Do not copy an example from documentation because release values change.

Use the shell selector

Bash uses \ line continuation. PowerShell uses the backtick. Select the shell you are actually using before copying.

Verify rollout and the first report

kubectl rollout status deployment/krevopilot-agent -n krevopilot-system
kubectl get pods -n krevopilot-system
kubectl logs deployment/krevopilot-agent -n krevopilot-system --tail=50

Connected

Logs contain Snapshot accepted id=.... The Clusters page becomes live and Overview fills after the first accepted report.

Not connected

A 401 means the key or saved connection identity is wrong. A timeout usually means outbound network access is blocked.

The first report usually appears within 60 seconds after a healthy rollout. Do not hard-refresh repeatedly; use Refresh on the cluster row and inspect the agent logs.

Reconnect an existing cluster

Reconnect the existing saved cluster record so optimization history and investigations remain attached.

Open Clusters and select the offline cluster.
Choose Fix connection or Reconnect agent.
Generate a new key only when the current key is unavailable or rejected.
Run the generated reconnect command in that exact cluster.
Wait for Snapshot accepted, then verify the cluster row is live.

Do not create a second cluster record for the same physical cluster. That splits history and makes data look duplicated.

Upgrade the agent and chart

When an update badge appears on Overview or Clusters, open the selected cluster and choose Upgrade agent/chart. Use the generated command; it preserves the saved key and current values while applying the supported release.

Upgrade conflict?

If Helm reports a conflict on .data.agent-key, do not delete the namespace or Secret blindly. Follow the safe ownership checks.

Uninstall

helm uninstall krevopilot-agent -n krevopilot-system

This stops future collection. Historical platform data follows the retention schedule until it expires or an authorized workspace deletion removes it. Delete the namespace separately only after confirming it contains no resources you intend to keep.