literature-review
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [NO_CODE]: The skill file consists entirely of instructional markdown and examples. It does not include any Python or Node.js scripts, shell commands, or configuration for external tool execution.
- [SAFE]: The skill references various research databases (arXiv, Semantic Scholar, PubMed, etc.) as part of its methodology. These are neutral references to established academic services for search strategy development and do not involve unauthorized network operations.
- [INDIRECT_PROMPT_INJECTION]: The workflow involves ingesting untrusted data from external research papers and preprints (SKILL.md). While there are no explicit boundary markers or sanitization rules provided for this data ingestion, the skill lacks an inventory of sensitive capabilities (such as file-system access, subprocess execution, or network exfiltration tools) that would be required to weaponize instructions embedded within the processed literature.
Audit Metadata