field
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of instructional markdown and reference files. No executable scripts (.py, .js, .sh) or binary files are present in the package, eliminating direct code execution risks.
- [SAFE]: Analysis of the instructions and reference guides reveals no hardcoded credentials, sensitive system paths, or malicious network exfiltration patterns. Data handling instructions explicitly prioritize participant privacy, PII anonymization, and compliance with SOC 2/GDPR standards.
- [SAFE]: The 'multi-engine' research recipe (Claude, Codex, Antigravity) is a well-structured orchestration pattern. While it involves processing external evidence through subagents, the system incorporates grounding checks for ethics, feasibility, and bias, and requires human review of the generated research plans, effectively mitigating indirect prompt injection risks.
Audit Metadata