presentations

Warn

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill relies on the 'Paper2All' project (https://github.com/YuhangChen1/Paper2All.git) for its Paper-to-Web/Video/Poster pipelines. This repository belongs to an unverified individual account and is not part of the trusted vendors list.
  • [REMOTE_CODE_EXECUTION]: Users are instructed to clone an external repository and run 'pip install -r requirements.txt'. This process involves downloading and installing unverified third-party dependencies, which could contain malicious code executed during the installation or runtime of the pipeline.
  • [COMMAND_EXECUTION]: Core workflows involve the execution of system-level commands, including 'google-chrome --headless' for PDF printing, 'ffmpeg' for video assembly, and 'libreoffice' for file conversion. These operations pose a risk if the input data (HTML, LaTeX, or PDF) is crafted to exploit these tools.
  • [PROMPT_INJECTION]: The skill's primary function is to ingest and transform untrusted external data, such as academic papers and PowerPoint templates. This creates a surface for indirect prompt injection, where instructions embedded within the processed research content could attempt to override the agent's behavior or exfiltrate information during the conversion process.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 9, 2026, 05:13 PM