Terminal-Bench-Science:在科学领域的研究工作流程中评估 AI 代理
Terminal-Bench-Science: Evaluating AI agents on research workflows across scientific domains