Operate Pipekit

Install the Pipekit Agent into your own Kubernetes cluster and connect it to Pipekit Cloud in 10 minutes.

A 10-minute path from "I have a Kubernetes cluster with Argo Workflows" to "Pipekit is managing it."

This quickstart assumes you're connecting your own cluster to Pipekit Cloud. For a fully self-hosted control plane (no Pipekit Cloud connection), see Self-Hosting Pipekit > Evaluate Self-Hosted instead.

Prerequisites

• A Kubernetes cluster with Argo Workflows installed.

• kubectl configured against that cluster.

• helm (v3) installed locally.

• A Pipekit Cloud account. If you don't have one, follow Evaluate Pipekit Cloud steps 1 to 2, then come back.

1. Create the Cluster in Pipekit

1. Log in to https://pipekit.io.

2. Decline the offer for a Pipekit-hosted free-trial cluster.

3. Click Connect A Cluster.

4. Select the Organization the cluster will belong to.

5. Give the Cluster a name (lowercase, no whitespace) and a description.

6. Click Submit.

Pipekit generates the install command and Helm values for your cluster.

2. Install the Pipekit Agent

Pipekit's UI prints a Helm install command of the form:

helm upgrade pipekit-agent oci://registry.pipekit.io/charts/pipekit-agent \
  --install \
  --namespace=argo \
  --set secrets.pipekitSecretAccessKey="<provided>" \
  --set secrets.pipekitClusterId="<provided>"

Run that command against your cluster. For the full chart values and configuration options, see Reference > Helm Chart Values > Agent.

3. Verify the agent is connected

kubectl get pods -n argo -l app=pipekit-agent

You should see a Pipekit Agent pod in Running state within a minute. The Pipekit UI's Cluster page also flips from Connecting to Active.

4. Submit a sample workflow

brew install pipekit/tap/cli   # or see other install methods in Reference > CLI
pipekit login
git clone https://github.com/pipekit/examples.git
cd examples
pipekit submit --cluster-name=<your-cluster-name> examples/dag-diamond/workflow.yaml

The CLI prints a link to the Run in the Pipekit UI. Open it to see the Run Graph, Pod Logs, and the submitted workflow YAML.

What's next

• Configure SSO and access control before inviting teammates: Access Control, SSO setup.

• Connect a Git provider so Workflows trigger on commits and pull requests: Pipes.

• Set up alerts for workflow failures: Alert Providers.

• Scale your cluster beyond the defaults: Operating at Scale.

If something goes wrong

The Pipekit Agent logs are your first stop:

kubectl logs -n argo -l app=pipekit-agent --tail=200

If the agent reports an auth or registration error, double-check the secret values from step 2. If it's reaching Pipekit but the Cluster still shows Connecting in the UI, check that the pipekitClusterId matches the Cluster you created in step 1.