codex

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

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill explicitly instructs the agent to use high-privilege sandbox modes and automatic execution flags (e.g., --sandbox workspace-write, --sandbox danger-full-access, --full-auto, --skip-git-repo-check) which enable writing, network/broad access, and automatic changes to the host environment — even though it recommends asking the user, the prompt still encourages actions that can modify the machine state.