cross-niche-outliers

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection by processing untrusted YouTube metadata in LLM prompts.
  • Ingestion points: Video titles and transcripts are fetched from YouTube and TubeLab in scripts/scrape_cross_niche_outliers.py and scripts/scrape_cross_niche_tubelab.py.
  • Boundary markers: Prompts in scripts/generate_title_variants.py use simple quotes to wrap titles but lack robust instructions to ignore embedded commands.
  • Capability inventory: The skill can execute subprocesses (yt-dlp), make network requests to multiple APIs, and write to Google Sheets.
  • Sanitization: No security-focused sanitization or filtering is applied to the retrieved video content before it is sent to the LLM.
  • [COMMAND_EXECUTION]: The script scripts/scrape_cross_niche_outliers.py executes the yt-dlp binary using the subprocess module. While arguments are currently derived from internal lists, the pattern of executing external binaries based on search results is a potential attack surface.
  • [EXTERNAL_DOWNLOADS]: The skill performs data retrieval from YouTube and uses APIs from TubeLab and Apify. These operations are essential for the skill's purpose and target established, well-known services.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 02:18 PM