openclaw-management

Warn

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill extensively uses the Bash tool to execute system commands for health monitoring, including curl for local gateway checks, pm2 for process management, and df/du for storage analysis.
  • [DYNAMIC_EXECUTION]: In the /openclaw-manage auto command, the skill dynamically writes a bash script to the file system at ~/clawd/tools/openclaw-doctor-quick.sh and subsequently executes it. It also uses python3 -c to execute inline code strings for processing JSON data from process managers.
  • [PERSISTENCE_MECHANISMS]: The skill attempts to maintain a continuous presence by scheduling its monitoring script via the openclaw cron command or instructing the user to manually modify the system crontab to run tasks every 15 minutes.
  • [PRIVILEGE_ESCALATION]: The skill modifies file permissions using chmod +x on the scripts it generates to ensure they can be executed by the system scheduler.
  • [DATA_EXFILTRATION]: The monitoring script includes logic to send status reports and alerts to an external Discord channel using the openclaw message send command. While intended for administrative alerts, this establishes a pathway for sending internal system state data to an external service.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 8, 2026, 02:29 PM
Security Audit — agent-trust-hub — openclaw-management