Node diagnostics

A rule-based health report for any node. Kubius reads its conditions, events, kubelet heartbeat and the pods scheduled on it, then surfaces what is actually wrong - ranked worst-first - so you are not eyeballing raw YAML during an incident.

• Flags OOM-killed pods and containers on the node
• Detects kubelet flapping, stale heartbeats and memory / disk / PID pressure
• Surfaces cordon state and control-plane taints
• Warns on pod-capacity strain and evicted pods
• Findings ranked critical-first, each linked to the underlying events