reverse-engineer-channel

Pass

Audited by Gen Agent Trust Hub on Nov 21, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content from YouTube, creating a surface for indirect prompt injection attacks.
  • Ingestion points: The agent is instructed to read external data including channel lists, video descriptions, and full transcripts (Steps 2, 9, and 10).
  • Boundary markers: There are no explicit instructions to use delimiters or to disregard instructions found within the transcripts or descriptions, which could allow an attacker to influence the agent's behavior via content in a video transcript.
  • Capability inventory: While the skill generates prompts for other tools and detailed reports, it does not demonstrate direct file system write access or arbitrary shell execution capabilities that could be exploited.
  • Sanitization: The instructions do not include steps for sanitizing or filtering the content retrieved from YouTube before the LLM processes it.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 21, 2026, 04:16 PM