industrial-ai-research
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a robust, multi-phase research workflow (Scope, Search, Collection, Verification, Synthesis, Assembly) with explicit safety boundaries and core rules that emphasize evidence-based reporting and source prioritization.
- [SAFE]: No unauthorized data access or exfiltration patterns were identified. Tools like WebSearch and WebFetch are strictly scoped to well-known academic and pre-print repositories (arXiv, IEEE Xplore) as defined in the source priority and venue map references.
- [SAFE]: No code obfuscation, dynamic execution, or hidden persistence mechanisms were found across the skill's instructions, modules, or reference files.
- [SAFE]: The 'survey-draft' workflow incorporates mandatory human checkpoints (outline approval) and quality gate checklists, reducing the risk of autonomous unauthorized actions or hallucinations.
- [SAFE]: Inter-skill communication, such as delegating LaTeX formatting to the 'latex-paper-en' skill, is handled through explicit user-approved handoffs as defined in the survey merge module.
Audit Metadata