universal-media-downloader
Fail
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes external command-line utilities including yt-dlp, ffmpeg, ffprobe, and aria2c via subprocess.run to handle media processing and downloading. It includes functionality to read browser cookies (Chrome, Firefox, Safari, Edge) automatically to authenticate with platforms like Douyin when default downloads fail.
- [EXTERNAL_DOWNLOADS]: The tool downloads media content and metadata from multiple third-party platforms such as YouTube, Bilibili, and Douyin to the user's Downloads directory. It uses network requests to fetch API responses and HTML content during the discovery and download process.
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it fetches and processes untrusted HTML and JSON data from external URLs provided by the user. This data is parsed using regular expressions and used to generate arguments for shell commands.
- Ingestion points: User-provided URLs and web content fetched from those URLs in scripts/download_media.py and scripts/download_douyin_video_nocookie.py.
- Boundary markers: Absent.
- Capability inventory: Shell command execution (yt-dlp, aria2c, ffprobe), network requests (requests), and file system writes.
- Sanitization: Relies on subprocess.run with argument lists; no explicit sanitization of external web content is performed before processing.
Recommendations
- HIGH: Downloads and executes remote code from: https://www.douyin.com/ - DO NOT USE without thorough review
Audit Metadata