multiline-validation

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's shipped Windows test YAMLs explicitly pull and run the external container image mcr.microsoft.com/powershell:lts-nanoserver-ltsc2022 at runtime, which executes remote code as a required dependency for the tests.

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 modify cluster/system state (apply a kube-system ConfigMap, restart daemonsets, perform helm upgrades, edit values.yaml, create/delete namespaces, and exec into pods to read system files), which requires elevated privileges and changes the machine/cluster state, so it must be flagged.