init
Warn
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands provided via the
--evalparameter. This is used to capture baseline metrics by running the command in the current working directory. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by interpolating external data (task descriptions and evaluation commands) into shell executions without sufficient sanitization.
- Ingestion points: The
{task}and{eval_cmd}parameters defined inSKILL.md. - Boundary markers: Absent. While arguments are wrapped in double quotes in the command template, this does not prevent shell command substitution or breakout techniques.
- Capability inventory: The skill can execute Python scripts and arbitrary shell commands in the local environment.
- Sanitization: No sanitization or validation logic for the input strings is mentioned or implemented in the instructions.
Audit Metadata