niche-bending

Pass

Audited by Gen Agent Trust Hub on Nov 21, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data from YouTube, including video titles, channel descriptions, and metadata. While an attacker could theoretically place malicious instructions in YouTube metadata, the skill's capabilities are limited to generating a markdown-based research report. There are no high-privilege tool executions (e.g., shell access, file writes) that could be subverted by such an injection, resulting in a low risk profile.
  • [EXTERNAL_DOWNLOADS]: The skill instructions specify the retrieval and display of YouTube video thumbnails (e.g., 'mqdefault.jpg'). These are standard media assets fetched from well-known YouTube content delivery networks (CDNs). This is a functional requirement for the skill's stated purpose and does not represent a security risk.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 21, 2026, 04:16 PM