Full threadrvnx·This shows the failure of human reviews alone, an LLM-based reviewer would have caught it. Both approaches are complementaryView on HN