Prefer using dedicated EKS Service Accounts

How it is checked

Audit method: Shell command

out=$(kubectl get pods -A -o go-template='{{range .items}}{{.metadata.namespace}} {{.metadata.name}} {{.spec.serviceAccountName}}{{"\n"}}{{end}}') || exit 1
bad=$(printf '%s\n' "$out" | awk -v ex='^(kube-system|kube-public|kube-node-lease)$' 'NF && $1 !~ ex && ($3=="default" || $3=="<no")')
if [ -z "$bad" ]; then echo compliant; else printf '%s\n' "$bad" | head -n 25; fi

Expected output

compliant

How it is fixed

Fix method: Guidance or infrastructure patch

Use dedicated Kubernetes service accounts for EKS workloads and associate AWS permissions to those service accounts instead of relying on the namespace default service account or broad worker node IAM permissions. Prefer EKS Pod Identity where supported, and use IRSA where Pod Identity is not supported or where IRSA is specifically required. Create one dedicated service account and one least-privilege IAM role for each application or distinct permission boundary, and update workload manifests to reference the intended service account explicitly. Disable automatic service account token mounting for workloads that do not need Kubernetes API access. After migrating workloads to dedicated service accounts, restrict access to the worker node instance profile so pods cannot inherit broader node credentials.

Check this host

Run this on a machine you are allowed to audit. The scan is read-only.

tmaegis scan --target eks