ml-paper-writing

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard shell commands including ls, find, grep, and cp for exploring research repositories and managing files. It also invokes LaTeX build tools such as latexmk and pdflatex, and the TeX Live package manager tlmgr to manage the writing environment.
  • [EXTERNAL_DOWNLOADS]: The skill fetches bibliographic data from well-known academic services like Semantic Scholar and the official DOI resolution service at doi.org. It also suggests installing TeX packages via tlmgr and recommends the Exa MCP server for real-time academic search.
  • [REMOTE_CODE_EXECUTION]: The instructions recommend adding the Exa MCP server using npx -y mcp-remote targeting https://mcp.exa.ai/mcp. This is a standard method for extending agent capabilities using well-known third-party academic tools.
  • [PROMPT_INJECTION]: The skill contains explicit safeguards against AI hallucination, instructing the agent to never generate BibTeX from memory and to use placeholders if verification fails, which promotes academic integrity.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 01:01 PM