research-paper-writing
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses a Makefile in the NeurIPS template to download style files from the official conference website (media.neurips.cc). This is an expected behavior for retrieving conference-specific formatting assets.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to search for and extract content from academic repositories like arXiv and Semantic Scholar. This presents a potential surface for indirect prompt injection if an external paper contains malicious instructions designed to influence the agent's behavior during synthesis or review. The skill mitigates this by emphasizing rigorous verification steps.
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands for LaTeX compilation (pdflatex, bibtex) and environment setup (apt install, pip install). These commands are well-scoped and essential for the stated purpose of producing scholarly manuscripts.
Audit Metadata