Kubernetes Scaling and Resilience Topology

When to use

Invoke when sizing autoscaling for a workload, hardening its survivability against node drains and zone loss, or auditing an inherited workload whose HPA/PDB/affinity is unsafe. This skill tunes and hardens the baseline HPA/PDB that k8s-workload-packaging-and-manifest created.

Do not use for: base Deployment/Service/HPA/PDB authoring (use k8s-workload-packaging-and-manifest); NetworkPolicy/RBAC (use k8s-network-and-identity-policy); the metrics pipeline that feeds autoscaling (use k8s-observability-and-operations-readiness); image hardening (use k8s-supply-chain-and-image-hardening); cluster/node provisioning and the cluster-autoscaler/Karpenter node layer (out of Family G — owned by the cloud platform stack and Terraform; this skill owns pod-level scaling only).

Inputs

Required:

• A workload manifest set from k8s-workload-packaging-and-manifest (the baseline HPA/PDB this skill tunes).
• Approved architecture/reliability decisions on SLOs, replica floors, and disruption budgets per tier, or explicit confirmation they are intentionally deferred.

Optional: