hypogenic
Pass
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads research datasets and source code from official Chicago Human+AI Lab (ChicagoHAI) GitHub repositories and installs the 'hypogenic' package from PyPI. These are recognized and reputable sources for research tools.
- [COMMAND_EXECUTION]: Documentation includes instructions for executing setup and processing scripts, such as 'setup_grobid.sh' and 'pdf_preprocess.py', which are standard components of the tool's scientific analysis workflow.
- [PROMPT_INJECTION]: The framework processes external dataset content by interpolating it into LLM prompt templates. While this represents an indirect prompt injection surface, it is necessary for the skill's primary function of data-driven hypothesis generation and is mitigated by the structured JSON input format.
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill uses environment variables for secure credential management and relies on well-known academic infrastructure.
Audit Metadata