harness-engineering
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is purely instructional and provides a framework for building secure autonomous systems. It explicitly details the use of 'Locked' surfaces for evaluation and 'Human-controlled' surfaces for destructive or irreversible actions, which are standard security mitigations against agent autonomy risks.
- [SAFE]: No malicious patterns such as prompt injection, data exfiltration, or unauthorized command execution were detected. The skill instructions focus on preventing 'metric gaming' and maintaining 'search discipline' through deterministic checks and human review.
- [SAFE]: External references point to well-known, legitimate research and software projects (e.g., Karpathy's autoresearch, HELM, AlphaEvolve) for educational purposes.
Audit Metadata