A new benchmark reveals AI coding agents excel at file discovery but falter at precise line identification. SWE-Explore highlights that insufficient context hinders agents from pinpointing the exact critical lines needed for code repair, impacting their overall effectiveness.
Opening Kapyn…