openai-on-aws

Plan agent experiments, benchmark real tasks, compare saved results, diagnose failures, audit graders, and explore charts and tool replays.