infrastructure-ops

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to detect password prompts and send {sudo_password} (and similar secret placeholders) into interactive sessions, which requires the LLM to handle and emit secret values verbatim into commands/inputs.

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 repeatedly instructs the agent to run privileged operations (sudo apt/upgrade, sudo certbot, systemctl restart, cp/sed on /etc/*, handle sudo password prompts and consider NOPASSWD), modify system configuration and service state, and thus actively pushes the agent to change and potentially compromise the host state.