kf-g-agent-review-adversarial-evaluation
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a logical framework for critical evaluation, establishing clear boundaries with other specialized review skills (e.g., post-implementation reviews). It provides specific output templates and procedures for identifying risks and proposing countermeasures without introducing any malicious patterns.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-supplied artifacts such as designs, plans, and proposals.
- Ingestion points: User-provided artifacts as defined in
SKILL.md. - Boundary markers: The output is strictly segmented into mandatory sections: Conclusion, Blocking issues, Non-blocking risks, Open questions, and Acceptance conditions.
- Capability inventory: No high-risk capabilities (network access, shell command execution, or file writing tools) are requested or utilized by this skill.
- Sanitization: While no explicit input sanitization is defined, the skill's adversarial posture is specifically designed to identify and highlight potential risks or contradictions in the input data, serving as a safety-oriented evaluation tool.
Audit Metadata