ResearchTasksMulti-Agent Collaboration Role OversteppingMulti-Agent Collaboration Role OversteppingFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedSWE total full set (dev)ChatDev0.2Overstepping Rate (<INFO>)4Apr 6, 2026SWE hard subset (dev)ChatDev0Overstepping Rate (<INFO>)4Apr 6, 2026SWE easy (dev)ChatDev0.4Overstepping Rate (<INFO>)4Apr 6, 2026