watching-videos
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The
watch-skill doctorcommand facilitates the automated download and installation of system dependencies, specificallyffmpegandyt-dlp.- [COMMAND_EXECUTION]: The skill uses theBashtool to executewatch-skillCLI commands, where user-supplied inputs such as URLs and local file paths are directly interpolated into the command string.- [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection because it processes and interprets untrusted text data extracted from video content. - Ingestion points: External URLs, local video files, OCR text extracted from video frames, and video transcripts in
SKILL.md. - Boundary markers: Absent. The instructions do not provide delimiters or specific warnings to the agent to ignore embedded instructions within the analyzed video content.
- Capability inventory: The skill possesses the ability to execute shell commands via
Bashand read local files. - Sanitization: No sanitization or validation logic is specified for the text content derived from OCR or transcripts before the agent processes it.
Audit Metadata