container-security-hardening

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

• Potentially malicious external URL detected (high risk: 0.90). The skill contains explicit runtime commands that download and execute remote install/manifests (e.g., sh of https://raw.githubusercontent.com/aquasecurity/trivy/main/contrib/install.sh and https://raw.githubusercontent.com/anchore/grype/main/install.sh, and kubectl apply of https://raw.githubusercontent.com/open-policy-agent/gatekeeper/release-3.17/deploy/gatekeeper.yaml, plus helm repo adds that fetch charts), meaning these URLs are fetched at runtime and run/installed, which can directly execute remote code.

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 require root on the host (e.g., "sudo apparmor_parser -r /etc/apparmor.d/docker-default") and other instructions that modify system state (apt install, useradd/adduser in build steps), which push the agent to perform privileged host changes.