yt-comment
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-generated content from YouTube comments, creating an attack surface for indirect prompt injection.\n
- Ingestion points: Untrusted data is ingested from external YouTube comment sections using the
list_repliestool.\n - Boundary markers: The instructions lack specific boundary markers or delimiters to isolate untrusted comment text from agent instructions during the triage and writing process.\n
- Capability inventory: The skill possesses write capabilities through tools such as
reply(to post comments) andhide_reply(to moderate comments).\n - Sanitization: The skill implements a mandatory human-in-the-loop (HITL) gate, requiring explicit user approval ('ship it') before any tool execution occurs, which significantly reduces the impact of potential injection attacks.
Audit Metadata