patsnap-lifescience-biomarker-investigation
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured instructions for pharmaceutical R&D tasks, utilizing specialized tools for patent and literature search.
- [PROMPT_INJECTION]: The instructions contain standard operational constraints and formatting requirements. No attempts to bypass safety filters or override system instructions were found.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill does not access sensitive local files or environment variables. All network-related activity is mediated through specific search and fetch tools for public scientific data.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including patent databases, academic literature, and web searches (SKILL.md). While this represents a surface for indirect prompt injection, it is necessary for the skill's primary function and no exploitable capability like arbitrary code execution was identified.
- [COMMAND_EXECUTION]: There are no shell commands, system calls, or privilege escalation attempts within the instructions.
Audit Metadata