arXivYuqiao Tan, Jinxiang Meng, Fangyu Lei, Minzheng Wang, Shizhu He, Jun Zhao, Kang LiuMon, Aug 3, 2026, 10:03 AM PDT

score 17.0

Coding agents fail when users edit code mid-task

Original: SWE-Touch: Benchmarking Coding Agents When Users Touch the Code

Writing ELI5 summary…