testing-skills-with-subagents
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of documentation and instructional frameworks for testing agent compliance. No malicious code or scripts are present.
- [PROMPT_INJECTION]: While the skill describes techniques for testing agent behavior (such as "pressure scenarios"), these are designed for documentation quality assurance and do not attempt to bypass system safety filters or override global constraints.
- [DATA_EXFILTRATION]: No network operations, credential access, or sensitive file paths were detected. The examples use generic or illustrative paths like
/tmp/payment-system. - [REMOTE_CODE_EXECUTION]: The skill does not download or execute external scripts. A reference to a local export script (
scripts/export-gemini-skill.py) is provided for documentation purposes but is not executed by the skill itself.
Audit Metadata