Find why your GKE workloads are not scaling.
Diagnoses GKE autoscaling failures and proposes manifest or configuration corrections.
When to use it
Use when an HPA does not scale a workload as expected or reports metric errors, not for authoring autoscaling configurations or sizing nodes.
Give it an HPA scaling problem; it investigates non-interactively and returns a reviewable correction or operator command without applying live changes.
Uses gcloud to fetch cluster credentials before live diagnostics; if live execution fails, it falls back to commands for the operator.
Uses kubectl to inspect the HPA's metrics, conditions, events, and manifest during live diagnostics.
This skill
GKE
Inspects your cluster's HPA
Cloud Logging
Reads historical autoscaling events