A Kubernetes cluster is a set of machines, called nodes, run by a control plane. You never log in to the nodes to start applications. Instead you send requests to the control plane's API server, and kubectl is the command-line client that does it.
kubectl get lists objects: kubectl get nodes shows the machines, kubectl get pods the running units of your applications. Most objects live in a namespace, a named area of the cluster. kubectl reads one namespace at a time, default unless you say otherwise, so an empty list can simply mean you looked in the wrong place. Add -n NAME for another namespace or -A for all of them.
Output options do much of the work: -o wide adds columns such as the node, -l filters by label and -o jsonpath prints a single field for a script. This course runs a three-node cluster on Kubernetes 1.37 inside your browser. It behaves like a real one, but nothing leaves the page.
From the whole cluster down to one field of one Pod.