arXiv · Software Engineering· Yiming Cheng (The University of Chicago), Alfin Wijaya Rahardja (Fudan University), Mengshi Zhang (TensorBlock, Inc), Zihao Chen (TensorBlock, Inc), Zhenpeng Chen (Tsinghua University), Yiling Lou (University of Illinois Urbana-Champaign)·· 7 小时前AI 评分42
AgentBug-Smith：自动复现 Agentic 系统中的真实 Harness Bug
AgentBug-Smith: Automatically Reproducing Real-World Harness Bugs in Agentic Systems
AI 导读
AgentBug-Smith 是一种自动复现真实 harness bug 的方法，从开源 agentic 系统中持续发现并复现 bug，在不同骨干 LLM 上比通用软件 bug 复现技术成功率高出 10.67%–27.56%。
来源：arXiv · Software Engineering · arxiv.org