subagent-testing
Pass
Audited by Gen Agent Trust Hub on May 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of documentation and best practices for testing AI agent skills. No executable malicious code or dangerous instructions were found across the provided files.
- [COMMAND_EXECUTION]: The skill includes shell command examples and Python code snippets within the markdown documentation. These are provided as templates for the user to follow during testing and are not automatically executed by the agent.
- [PROMPT_INJECTION]: The skill discusses techniques to detect and mitigate 'rationalization' (a form of prompt bypass), but does not contain any instructions meant to override agent safety or system prompts.
Audit Metadata