hamodywe

Hhamodywe/unfinished

Finds the LLM calls that use the answer without checking whether the model finished giving it — the truncation that looks like a flaky model and bills like a retry loop.

Hhamodywe/unscaled

Finds the Kubernetes autoscalers that will never add a pod — the HPA whose target has no CPU request, the ScaledObject pointing at a renamed workload, the replicas field fighting it every sync.

Hhamodywe/ghosttest

Counts the tests in your repository, then counts the ones that actually run — and tells you where the difference went.