watching-videos

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content from video URLs and local files, including OCR text and transcripts. This creates an attack surface where instructions embedded in the video's audio, visual text, or metadata could attempt to influence the agent's behavior.
  • [EXTERNAL_DOWNLOADS]: The watch-skill doctor command is documented to install external system dependencies including ffmpeg and yt-dlp if they are missing from the environment.
  • [COMMAND_EXECUTION]: The skill relies on executing shell-based commands (watch-skill) to perform video analysis, indexing, and search operations on the local filesystem.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 09:03 AM