PhoenixRepair is a multi-agent framework for improving automated software issue resolution by expanding repair-strategy exploration. It samples multiple candidate edit locations, optionally uses graph-based localization for difficult tasks, and applies iterative reflection and refinement to patch generation. A final generation round is guided by distilled insights from prior attempts. On SWE-bench-Verified, the authors report a 7.8% relative improvement over SWE-agent when using DeepSeek-V3.1, and a 76.0% Pass@1 resolved rate with MiniMax-M2.5. The framework also reports higher fault-localization accuracy than existing approaches, with code released on GitHub.
No heat snapshots are available in the last 24 hours.