academic-paper

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [SAFE]: Extensive review of all 45 files, including agent roles and reference guides, confirms that the skill is safe for use. There are no attempts to access sensitive system files, hardcode credentials, or bypass agent safety protocols. All external URLs point to legitimate academic or government institutions.
  • [COMMAND_EXECUTION]: The formatter_agent generates shell commands to facilitate the conversion of manuscripts into LaTeX, DOCX, and PDF formats. These commands use standard tools like Pandoc and Tectonic and are provided for user-directed execution.
  • [PROMPT_INJECTION]: The skill's instructions were evaluated for injection patterns; none were detected. The agents follow structured protocols that maintain professional academic personas and prioritize user guidance through Socratic dialogue.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided research materials and reviewer comments. While these are external inputs, the analysis found no evidence of malicious instructions being processed or resulting in unintended agent actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 03:11 AM
Security Audit — agent-trust-hub — academic-paper