scientific-thinking-scholar-evaluation
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of natural language instructions for evaluating scientific artifacts and does not include any executable code, shell commands, or scripts.
- [SAFE]: No patterns related to prompt injection, data exfiltration, obfuscation, or privilege escalation were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external documents (papers, proposals, and literature reviews). While these are untrusted data sources, the skill lacks any tool-calling capabilities, file system write access, or network access that could be exploited by embedded instructions, making the risk negligible and consistent with its primary purpose.
Audit Metadata