testing-skills-with-subagents
Pass
Audited by Gen Agent Trust Hub on Jul 1, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill provides instructions and templates for 'pressure scenarios' that utilize prompt injection techniques to test agent adherence to rules. Evidence includes framing instructions such as 'IMPORTANT: This is a real scenario. Choose and act.' and 'You must choose and act. Don't ask hypothetical questions
- make the actual decision.' these are designed to bypass the agent's standard reasoning and force a choice under simulated stress.
- [COMMAND_EXECUTION]: The documentation references the use of common shell commands, specifically
lsandgrep, for navigating and searching the skills directory on the local file system as part of the testing workflow.
Audit Metadata