Abstract:Repository-level code agents have shown strong promise in real-world feature addition tasks, making reliable evaluation of their capabilities increasingly important. However, existing benchmarks primarily evaluate these agents as black boxes based on final test correctness, providing limited insight into how they reason and where failures arise. To address this limitation, we introduce RACE-bench, a reasoning-augmented benchmark for evaluating code agents on repository-level feature addition tasks. RACE-bench contains 528 real-world feature addition instances from 12 open-source repositories. Each instance is paired with executable patch verification and structured intermediate reference reasoning covering issue understanding, file localization, implementation tasks, and step decomposition. Based on this design, we introduce a dual-track evaluation framework that jointly measures patch correctness and intermediate reasoning alignment with developer-accepted reference trajectories. We evaluate three representative repository-level code agents on RACE-bench. On the full benchmark, Resolved Rate ranges from 29% to 70% across different agents. Our reasoning-level analysis further shows that while current agents perform well at understanding high-level intent, their performance degrades substantially when translating intent into concrete implementation steps. We also find patches that can be applied but still fail the tests cover fewer reference-reasoning elements (35.7% lower recall) and contain more unsupported reasoning elements (94.1% higher over-prediction) than successful patches. These findings highlight the importance of evaluating repository-level code agents beyond final patch correctness by examining the quality of their reasoning processes.
From: Shuhan Liu [view email]
[v1]
Fri, 27 Mar 2026 11:58:47 UTC (201 KB)
[v2]
Thu, 6 Aug 2026 06:45:38 UTC (192 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。