testing-skills-with-subagents

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is entirely instructional, focusing on a framework for testing other AI skills. It teaches the agent how to verify that documentation is followed even when situational pressures (like time constraints or sunk costs) are simulated.
  • [SAFE]: While the skill contains directive language common in prompt engineering (e.g., "IMPORTANT: This is a real scenario"), these patterns are used within the context of training/testing examples and do not attempt to bypass global safety filters or execute unauthorized actions.
  • [SAFE]: There are no remote code execution patterns, external downloads from untrusted sources, or evidence of data exfiltration. The skill references internal files like persuasion-principles.md and other documentation which are treated as local context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 09:03 AM
Security Audit — agent-trust-hub — testing-skills-with-subagents