How do I verify if the autoscaler policies are active?

The skill includes diagnostic commands to query the CAST AI API and check the lifecycle distribution (spot vs. on-demand) of your managed nodes.

What are the prerequisites for using this CAST AI skill?

You need a CAST AI API Key and Cluster ID, and your cluster must have the CAST AI controller and evictor installed (Phase 2 automation).

Does this skill support Terraform?

Yes, it includes specific patterns and instructions for creating and managing node templates using Terraform HCL for infrastructure-as-code workflows.

How does this skill help reduce Kubernetes costs?

It automates the configuration of CAST AI policies that prioritize spot instances, scales down empty nodes automatically, and manages resource headroom efficiently.

Can I use this for GPU-specific workloads?

Absolutely. The skill provides templates for configuring specialized nodes, such as GPU-enabled on-demand instances, alongside standard spot-worker pools.

CAST AI Autoscaler & Policy Optimizer

Name: CAST AI Autoscaler & Policy Optimizer
Author: jeremylongshore

byjeremylongshore

•

2,028

•

Infraestructura en la Nube

Automates the configuration of CAST AI autoscaler policies and node templates to optimize Kubernetes cluster costs.

This skill empowers developers and DevOps engineers to streamline Kubernetes cost management by configuring CAST AI's advanced autoscaling features directly within Claude Code. It facilitates the setup of spot instance policies, node downscaler settings, and evictor configurations through both direct API interactions and Terraform resource management. By implementing automated cost-optimization patterns, it ensures clusters remain highly available while minimizing cloud spend through intelligent workload placement and resource lifecycle management.

Características Principales

01Enable node downscaler and evictor for automated resource cleanup

02Configure spot instance diversity and price increase limits

03Set cluster-wide CPU and memory headroom for unschedulable pods

04Real-time verification of node lifecycle status and managed counts

052,028 GitHub stars

06Generate Terraform HCL for workload-specific node templates

Casos de Uso

01Transitioning a standard EKS/GKE cluster to automated cost-optimization using spot instances

02Tuning autoscaler policies to balance infrastructure performance with aggressive cost savings

03Defining custom node templates for specialized workloads like GPU processing or batch jobs

Características Principales

01Enable node downscaler and evictor for automated resource cleanup

02Configure spot instance diversity and price increase limits

03Set cluster-wide CPU and memory headroom for unschedulable pods

04Real-time verification of node lifecycle status and managed counts

052,028 GitHub stars

06Generate Terraform HCL for workload-specific node templates

Casos de Uso

01Transitioning a standard EKS/GKE cluster to automated cost-optimization using spot instances

02Tuning autoscaler policies to balance infrastructure performance with aggressive cost savings

03Defining custom node templates for specialized workloads like GPU processing or batch jobs