We built a benchmark that teaches AI agents not to get scammed. Ask us anything!