When we assess a Kubernetes cluster, we rarely begin inside the pods. The most productive first steps are the public surface of the control plane, exposed dashboards and the permissions attached to default accounts.
A single unauthenticated endpoint or an over-scoped service account is often enough to pivot from a throwaway pod to cluster-admin. The finding is usually the same: a default that was never revisited.
Hardening does not require exotic tooling. It requires knowing where the defaults live, and treating every one of them as a decision rather than a given.