youtube-research
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data retrieved from YouTube transcripts and video metadata via the
yt-insightstoolset. - Ingestion points: The agent ingests untrusted text via the
search_passagesandget_passagetools as described inSKILL.md. - Capability inventory: The available tools are restricted to read-only search and retrieval functions (
list_corpora,search_videos,search_passages,get_passage). - Boundary markers: The instructions command the agent to distinguish direct evidence from interpretation, but do not provide explicit delimiters or instructions to ignore embedded commands within the transcripts.
- Sanitization: There is no explicit mention of filtering or sanitizing the transcript content before it is processed by the agent.
Audit Metadata