Show HN: A replayable A2A jury for tracing how agents influence decisions

A developer has created an interactive simulation called "AI Liability Tribunal" that places autonomous agents with different roles and incentives into a fictional legal case involving a robotaxi fatality, allowing users to observe how agents communicate and influence each other's decisions through a replayable interface. The system produces JSON results, traces, transcripts, and interactive HTML reports to demonstrate agent-to-agent communication dynamics, with jurors able to choose whom to question and track opinion changes after each message. The project is explicitly framed as a software experiment for studying multi-agent interaction patterns, not as legal analysis or safety validation.