Tag
A new open source tool enables local debugging of AI agents by viewing traces, allowing automated eval writing and testing with tools like codex and Claude code.