Testing My AI Agent Against Red Team Attacks,100 scenarios, security validation and Qwen3 evaluation.