detecting-container-escape-attempts

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's Step 1 uses "helm repo add falcosecurity https://falcosecurity.github.io/charts" and "helm install falco falcosecurity/falco", which fetches remote Helm chart content at runtime and deploys/executes that remote code/manifests (a required dependency for the skill).

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 instructs privileged, system-level actions (installing kernel/eBPF probes, writing /etc/falco and /etc/audit rules, applying seccomp profiles, helm installs, and even an "Exploitation" step) that require root and modify host state, so it pushes the agent to perform potentially compromising changes to the machine.