youtube
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the execution of shell commands including
uvx,yt-dlp,sed,awk, andjq. These commands are used to manage downloads and process text files, involving write operations to user directories such as~/Downloads,~/Music, and specific synchronization folders. - [EXTERNAL_DOWNLOADS]: The skill uses
uvxto dynamically fetch and execute theyt-dlppackage from the Python Package Index (PyPI) at runtime. - [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection through its processing of external data.
- Ingestion points: Video metadata (JSON format), titles, descriptions, and transcript files (
.srt,.vtt) retrieved from YouTube. - Boundary markers: Absent; the skill does not implement delimiters or specific instructions to the agent to ignore embedded instructions in the downloaded content.
- Capability inventory: Shell subprocess execution, file system writes across multiple user-home subdirectories, and network access to YouTube services.
- Sanitization: Absent; while the skill uses
sedto remove timestamps from transcripts, it does not perform sanitization or validation of the text content against malicious instructions.
Audit Metadata