Tag
The paper presents a new benchmark for evaluating coding agents that assesses their ability to complete real-world customer service tasks in a realistic environment.