EKS clusters without any compute

What LeftSize checks

Find active EKS clusters (older than 7 days) with no compute: no managed node groups, no Fargate profiles and no running EC2 instances tagged for the cluster (covers self-managed node groups and Karpenter). EKS Auto Mode clusters are skipped.

Why it matters

Resources like this keep billing every hour while delivering little or no value, and nobody notices until the invoice arrives.

What you get in the GitHub issue

LeftSize runs this check from a scheduled GitHub Action with a read-only identity in your AWS account and opens one issue per scope, updated on every scan. The issue includes:

• The affected resources, grouped per subscription, resource group or account and region
• An estimated monthly cost where pricing data is available
• Background on why it was flagged and when it is a false positive
• Step-by-step remediation with safety checks
• Ready-to-run fix scripts (CLI, PowerShell or Terraform where available)
• Cost and operational impact of fixing it
• Alternative approaches and their trade-offs
• Snooze, ignore or report a finding with a comment on the issue