A multi-agent benchmark for LLM deliberation, simulating a cinematic jury environment to evaluate persuasion, stubbornness, and decision-making dynamics. - View it on GitHub