Close-reading workbench for LLM experiment outputs — JSONL/CSV/inspect-ai .eval viewer with marks, judges, SQL, and an embedded Claude session
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).