youtube-seo
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
ffmpeginscripts/extract-audio.shto extract audio from video files provided by the user. This is a standard media processing operation and is consistent with the skill's primary purpose of preparing content for transcription. - [REMOTE_CODE_EXECUTION]: The workflow mentions executing a transcription script (
transcribe.sh) located in a relative path{skillsDir}/openai-whisper-api/. This references a local dependency within the agent environment rather than downloading or executing untrusted code from the internet. - [DATA_EXFILTRATION]: No unauthorized network operations or exfiltration patterns were found. The skill suggests a community link to Telegram (
t.me/YOUR_CHANNEL), which is standard for social media marketing metadata. - [PROMPT_INJECTION]: The
SKILL.MDfile contains formatting rules and style guides (e.g., 'Do not use long dashes', 'Max 1 emoji'). These are legitimate formatting constraints for a copywriter-style agent and do not attempt to bypass core AI safety filters or hijack agent behavior.
Audit Metadata