youtube-ideation

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted data from external research sources such as YouTube Studio Trends, Google Trends, and niche competitor analysis. This ingestion creates a surface for indirect prompt injection where malicious instructions embedded in external content could influence agent behavior. Ingestion points include 02-DOCS/*.md and external research data described in SKILL.md. No specific boundary markers are defined for the interpolation of this data into prompts.\n- [COMMAND_EXECUTION]: The skill includes and instructs the execution of a local bash script, scripts/verify.sh, to validate the structural integrity of its generated markdown artifacts. This script performs read-only parsing using awk and does not involve network operations or privilege escalation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 09:37 PM
Security Audit — agent-trust-hub — youtube-ideation