← All field notes

What a CIS Kubernetes Benchmark Audit Actually Checks

What a CIS Kubernetes audit checks across the control plane, etcd, worker nodes and security policies—and how to turn findings into a practical remediation plan.

CIS Kubernetes audit checklist covering the control plane, etcd, kubelet and security policies.

A cluster isn't secure or insecure in the abstract. It is compliant or non-compliant against a specific set of controls. The CIS Kubernetes Benchmark is the most widely used baseline for that, and it is worth knowing what it inspects before you commission or run an audit.

What the CIS Kubernetes Benchmark is

The Center for Internet Security (CIS) publishes benchmarks: detailed, consensus-built configuration checklists. The Kubernetes benchmark groups its checks by component and marks each as automated or manual.

Managed services have their own versions, such as the CIS Amazon EKS, Google GKE and Azure AKS benchmarks. These skip the parts the cloud provider controls and focus on what you still own.

The four areas a CIS audit checks

1. Control plane components

The audit checks how the API server, controller manager and scheduler are configured. Typical findings:

  • Anonymous authentication left enabled on the API server.

  • Audit logging not configured, so there is no record of who changed what.

  • Admission controllers missing or misconfigured.

Many clusters still run defaults that were reasonable years ago but no longer are.

2. etcd

etcd stores every Secret in the cluster, so it gets close attention:

  • Is encryption at rest configured?

  • Do clients and peers authenticate with certificates?

  • Is etcd reachable from outside the control plane network?

3. Worker node configuration

The kubelet on every node is checked individually:

  • Anonymous authentication on the kubelet API.

  • The authorization mode (it should not allow every request).

  • The read-only port, which exposes pod information without authentication.

On EKS, GKE and AKS much of this is handled for you. On self-managed or on-prem clusters such as kubeadm or RKE2, this risk is entirely yours.

4. Policies: RBAC, Pod Security and network

This is where most of the real backlog sits, and where standard dashboards show nothing:

  • Wildcard RBAC roles and unnecessary cluster-admin bindings.

  • Service accounts with far more rights than they use.

  • Namespaces with no NetworkPolicy, so every pod can reach every other pod.

  • Privileged containers or pods running as root.

How to run a first check yourself

kube-bench, an open-source tool from Aqua Security, runs the CIS checks as a Kubernetes Job. Start it with:

bash
kubectl apply -f https://raw.githubusercontent.com/aquasecurity/kube-bench/main/job.yaml

Then read the results:

bash
kubectl logs job/kube-bench

Each check comes back as PASS, FAIL or WARN, with a remediation hint for every failure.

What a CIS score actually means

A CIS-aligned score is a point-in-time snapshot of how much of a defined, severity-weighted control set you satisfy. The number on its own matters less than what sits behind it: a prioritised backlog of exactly what to fix, in order of risk.

That backlog is what makes remediation predictable. You know the scope, so the work can be planned and priced instead of left open-ended.

Why we keep Audit and Harden separate

At Turf Dev, Audit and Harden are separate engagements, on purpose:

  • Independence: an auditor who also fixes the findings can't credibly mark their own work.

  • Clear scope: hardening against a scored backlog has a defined end. Hardening without one never finishes.

  • Proof: a rescan after hardening shows the score moved, with evidence.

How Turf Dev can help

Turf Dev's Kubernetes audit maps your cluster against the CIS Benchmark, adds the policy review that automated tools miss, and hands you a severity-ranked fix list. When you are ready, we harden against it and rescan to prove the result. Reach out at turfdev.in.

Discussion

Have a question or something to add? Share it below.

Comments appear immediately. Please keep the discussion respectful.

No comments yet. Be the first to join the discussion.

Keep exploring

More from the field.

All articles ↗

Platform Mandate

Request Architecture Review

Review platform posture, reliability constraints, and cost governance priorities with an infrastructure engineering team.

Request Architecture Review