Agent Skills: Performing Container Escape Detection

>

UncategorizedID: autohandai/community-skills/performing-container-escape-detection

Install this agent skill to your local

pnpm dlx add-skill https://github.com/autohandai/community-skills/tree/HEAD/performing-container-escape-detection

Skill Files

Browse the full folder contents for performing-container-escape-detection.

Download Skill

Loading file tree…

performing-container-escape-detection/SKILL.md

Skill Metadata

Name
performing-container-escape-detection
Description
>

Performing Container Escape Detection

Instructions

Audit Kubernetes pods for container escape vectors including privileged mode, dangerous capabilities, host namespace sharing, and writable hostPath mounts.

from kubernetes import client, config
config.load_kube_config()
v1 = client.CoreV1Api()

pods = v1.list_pod_for_all_namespaces()
for pod in pods.items:
    for container in pod.spec.containers:
        sc = container.security_context
        if sc and sc.privileged:
            print(f"PRIVILEGED: {pod.metadata.namespace}/{pod.metadata.name}")

Key escape vectors:

  1. Privileged containers (full host access)
  2. CAP_SYS_ADMIN capability
  3. Host PID/Network/IPC namespace sharing
  4. Writable hostPath mounts to / or /etc
  5. Docker socket mount (/var/run/docker.sock)

Examples

# Check for docker socket mounts
for vol in pod.spec.volumes or []:
    if vol.host_path and "docker.sock" in (vol.host_path.path or ""):
        print(f"Docker socket exposed: {pod.metadata.name}")