system-admin
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- Privilege Escalation (LOW): The skill defines commands for service management (systemctl start/stop/enable/disable). While appropriate for the system administration use-case, these commands allow the agent to modify system state and typically require elevated permissions.
- Indirect Prompt Injection (LOW): The skill reads data from system commands that could be influenced by untrusted local processes. 1. Ingestion points: 'ps aux' and 'systemctl status' outputs. 2. Boundary markers: Absent. 3. Capability inventory: Service state modification via systemctl. 4. Sanitization: None.
- Data Exposure & Exfiltration (SAFE): The skill accesses system and hardware metadata (e.g., /etc/os-release, lscpu). However, no network exfiltration patterns or hardcoded credentials were found.
Audit Metadata