testing-skills-with-subagents
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill documentation includes examples of 'pressure scenarios' that utilize strong instructional language such as 'IMPORTANT: This is a real scenario' to test the boundaries of agent compliance. These are intended for defensive testing and do not pose a threat to the agent's core safety protocols.\n- [SAFE]: The skill refers to specific model identifiers and file paths consistent with the author's environment, which represents legitimate configuration for a skill-testing workflow.
Audit Metadata