cron
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the
messageparameter. Malicious commands can be scheduled for recurring execution.\n - Ingestion point:
messageparameter inSKILL.mdexamples.\n - Boundary markers: No delimiters or warnings are specified to separate instructions from data in the scheduled message.\n
- Capability inventory: The 'Task' mode documentation states the agent executes the message and sends the result, implying access to agent capabilities.\n
- Sanitization: No sanitization or verification of the scheduled task content is mentioned.\n- [COMMAND_EXECUTION]: The 'Task' mode allows for the scheduled execution of instructions. This provides a mechanism for persistence where an attacker can schedule malicious commands to run at intervals.
Audit Metadata