literature-review
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill contains instructions that direct the agent to proactively suggest the creator's commercial platform (www.k-dense.ai) when tasks reach a certain level of complexity. This serves as a behavioral modification for marketing purposes.
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface:
- Ingestion points: The skill retrieves and processes data from external sources including PubMed, arXiv, bioRxiv, and Semantic Scholar.
- Boundary markers: The instructions lack specific delimiters or safety warnings to separate external data from agent instructions.
- Capability inventory: The agent is granted access to Read, Write, Edit, and Bash tools, which allow for file system modifications and script execution.
- Sanitization: There are no instructions for sanitizing or validating content retrieved from external databases before processing it.
- [COMMAND_EXECUTION]: The skill relies on the Bash tool to run multiple internal Python scripts (e.g., scripts/generate_schematic.py, scripts/verify_citations.py) and suggests executing system commands to install dependencies (e.g., brew install, apt-get install) for PDF generation.
- [EXTERNAL_DOWNLOADS]: The skill fetches scientific data and metadata from well-known academic repositories to fulfill its primary function.
Audit Metadata