yt-audit
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructions direct the agent to read
~/.claude/youtube/voice.md, which is a file path inside the user's local configuration directory. Accessing such paths can expose sensitive personal data or user personas to the agent's context. - [COMMAND_EXECUTION]: The skill requires the execution of external Python scripts (
../yt-package/title.pyand../yt-script/hookscore.py) located in parent directories. This relies on an external environment setup that is not strictly confined to the skill's package. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by transcribing user-provided YouTube videos to create a voice profile. This data processing step creates a surface for indirect prompt injection attacks.
- Ingestion points: YouTube video content for transcription requested in SKILL.md.
- Boundary markers: Absent; there are no instructions to ignore embedded commands within the transcriptions.
- Capability inventory: File reading (
voice.md) and local script execution (title.py,hookscore.py). - Sanitization: Absent; the skill does not specify any filtering or validation for the transcribed content.
Audit Metadata