hunt-k8s

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly instructs reading service-account tokens and Secret values (e.g., cat /var/run/secrets/.../token, decoding base64 secrets, etc.) and embedding them into Authorization headers or report output, which requires the agent/LLM to handle and potentially output secret values verbatim.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This document contains explicit, actionable offensive instructions (curl/etcdctl/docker.sock commands, token exfiltration, kubelet/API-mediated exec, privileged container creation, and CVE exploit guidance) enabling credential theft, remote code execution, and host compromise if used maliciously.

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 contains explicit commands that create privileged containers (bind-mounting host /), invoke kubectl debug/ephemeral privileged containers, and directly use the local docker socket—actions that would modify or escalate privileges on the machine hosting the agent.