yt-dlp
Fail
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads the
yt-dlpbinary from its official GitHub repository releases page (https://github.com/yt-dlp/yt-dlp/releases/latest/download/yt-dlp) if the tool is not already present on the system. - [REMOTE_CODE_EXECUTION]: The script downloads a binary from a remote source, grants it execution permissions using
chmod +x, and executes it in the shell to perform its core functions. - [COMMAND_EXECUTION]: The skill uses shell commands to invoke the
yt-dlpbinary with several parameters to download audio, video, and extract metadata like titles and descriptions. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill retrieves and prints video titles, descriptions, and subtitle files (
.vtt) from external YouTube URLs. - Boundary markers: No delimiters or isolation warnings are used when the agent reads the extracted metadata or subtitles.
- Capability inventory: The skill has shell execution capabilities and file system access to the
/tmp/directory. - Sanitization: There is no filtering or validation of the text content fetched from YouTube before it is ingested into the agent context.
Recommendations
- HIGH: Downloads and executes remote code from: https://github.com/yt-dlp/yt-dlp/releases/latest/download/yt-dlp - DO NOT USE without thorough review
Audit Metadata