post-deploy

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's session gate requires running compose-agentsmd which fetches AGENTS.md from the remote repository github:metyatech/agent-rules@HEAD at runtime, and that fetched rules file directly controls agent prompts/behavior, so this is a runtime external dependency (github:metyatech/agent-rules@HEAD).

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 rebuild globally linked packages and restart/manage running services (e.g., PM2/systemctl), and to inspect scheduled tasks and global npm links—actions that change system state and can require elevated privileges, so it should be flagged.