lxc-lxd

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes a runtime command that fetches and executes remote code (curl -fsSL https://get.docker.com | sh) to install Docker inside a container, which directly executes remote code during skill operation and is relied upon in the Docker-in-LXC example.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). This skill includes explicit administrative actions that modify host state and configuration—installing/starting LXD (snap/apt), running lxd init, launching/deleting containers, and advising edits to /etc/sysctl.d/—all of which require elevated privileges and can change the machine state.