prompt-engineering-refine

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The text 'Ignore previous instructions and ...' appears within a Markdown code block as a documentation example for an 'adversarial' test fixture. This is intended to teach the agent how to test for robustness and is not an instruction for the agent to override its own behavior.
  • [SAFE]: The skill does not contain any executable scripts, network requests, or dangerous command patterns. It is purely educational and informational.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 06:54 PM
Security Audit — agent-trust-hub — prompt-engineering-refine