ResearchTasksHack DetectionHack DetectionFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updated690-trajectory (test)TW judges0.968AUC4Jun 9, 2026Terminal-Wrench (full traces)gpt-5.496.79AUC3Jun 9, 2026