ml-paper-writing
Pass
Audited by Gen Agent Trust Hub on Mar 31, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Downloads official LaTeX style packages from the NeurIPS media server (media.neurips.cc) and references established research APIs like Semantic Scholar and CrossRef.
- [COMMAND_EXECUTION]: Employs standard system tools for file exploration (ls, find, grep) and academic document processing (pdflatex, bibtex, latexmk, pdfcrop).
- [PROMPT_INJECTION]: The skill ingests user-provided research repository data, creating an indirect prompt injection surface. This is mitigated by explicit instructions for the model to verify all citations programmatically rather than relying on ingested or generated text.
Audit Metadata