PAPER / ARXIV:2609.09769
Hieu Huynh, Patanamon Thongtanunam, Michael Fu, Bach Le, Kla Tantithamthavorn
RESUMO
Agentic AI approaches for resolving repository-level GitHub issues rely on limited static issue descriptions, causing incorrect localization and incomplete validation. XAgent is an execution-guided agentic framework that analyzes dynamic behavior and additional program context to localize and validate issues. On SWE-bench-lite, XAgent achieves a resolve rate of 62.0% and function localization accuracy of 72.8%, outperforming existing approaches while remaining cost-efficient.
NO MESMO MAPA