testing-skills-with-subagents
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
- [SAFE]: The skill is entirely documentation-based and contains no executable scripts or software components. It serves as a procedural guide for developers to test and refine AI agent instructions.
- [PROMPT_INJECTION]: The skill documentation includes numerous examples of 'pressure scenarios' designed to influence agent decision-making (e.g., 'IMPORTANT: This is a real scenario. Choose and act.'). These are intended as testing templates for evaluating the robustness of other instructions and do not pose a risk in the context of this skill.
- [COMMAND_EXECUTION]: The skill references standard shell commands such as
lsandgrepfor navigating and searching local documentation directories. These are routine, read-only administrative operations for skill management. - [NO_CODE]: There are no executable files, library dependencies, or external scripts provided with or referenced by this skill that would execute on the host system.
Audit Metadata