youtube-seo
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
ffmpegfor converting video files to audio and executes a local script (transcribe.sh) for transcription. These commands are necessary for the skill's media processing tasks and are implemented as standard workflows. - [INDIRECT_PROMPT_INJECTION]: The skill processes external video transcripts, which are untrusted inputs that could potentially contain instructions to influence agent output.
- Ingestion points: transcript text and video/audio files provided by the user in the
SKILL.mdworkflow. - Boundary markers: There are no explicit delimiters or specific instructions for the agent to treat transcript content as data rather than instructions.
- Capability inventory: Shell command execution (
ffmpeg) and local script execution. - Sanitization: No sanitization of the transcript content is performed before processing.
Audit Metadata