PodResourceAutoscaler (PRA) continuously adjusts container requests and limits based on real pod usage.

You can target a specific workload (`deployment`, `daemonset`, `statefulset`) or select pods by label selector.

Policy settings define polling cadence, cooldown behavior, and utilization thresholds for CPU and memory requests or limits.

Bounds are optional guardrails for min/max values and step changes, helping prevent oversized or undersized updates.