cron-scheduling

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides comprehensive instructions for executing arbitrary commands via crontab, systemctl, and at.
  • [COMMAND_EXECUTION]: Includes a Bash wrapper script (cron-wrapper.sh) that executes commands provided as arguments.
  • [COMMAND_EXECUTION]: Documents methods for performing privileged operations using sudo, such as managing system-wide timers and other users' scheduled tasks.
  • [PROMPT_INJECTION]: Debugging instructions involve reading system logs (syslog, journalctl), which presents an indirect prompt injection surface where malicious log content could influence agent behavior.
  • Ingestion points: System log files (/var/log/syslog, /var/log/cron) and journalctl output (SKILL.md).
  • Boundary markers: No markers or instructions to ignore embedded commands are included when reading logs.
  • Capability inventory: Full subprocess execution via scheduling tools like crontab and at (SKILL.md).
  • Sanitization: No sanitization or validation of log content is implemented before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 04:01 PM