benchmarking-kubernetes-with-kube-bench

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill contains commands that fetch and run remote code/manifests at runtime (e.g., the kube-bench release tarball "https://github.com/aquasecurity/kube-bench/releases/download/v${KB_VERSION}/kube-bench_${KB_VERSION}_linux_amd64.tar.gz" and the in-cluster job manifest "https://raw.githubusercontent.com/aquasecurity/kube-bench/main/job.yaml"), which are required dependencies and would execute code/configuration when applied.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly instructs the agent to use sudo and modify system files (e.g., moving binaries into /usr/local/bin, copying to /etc, editing /etc/kubernetes manifests, changing file permissions) and to run cluster-altering commands, which directly change the machine/cluster state and require elevated privileges.