Standalone execution trace that records every agent position, critique, revision, and convergence point across an entire Multi-Agent Debate reasoning session for auditability and continuous improvement.
