llm-evals-and-retrieval-quality
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill mentions common prompt injection strings such as "ignore previous instructions" and "jailbreaks". However, these are provided as examples of adversarial test cases that developers should include in their evaluation sets to improve system robustness. These patterns are used in a descriptive, educational context rather than as active instructions to override agent behavior.
- [SAFE]: No executable code, shell commands, or network operations are present in this skill. It functions as a documentation resource for AI engineering best practices.
Audit Metadata