Software engineering agents increasingly use reusable skills distilled from prior experience to resolve repository-level issues, yet such skills often fail to transfer across repositories. A central challenge is that a behavior appearing in a successful trajectory is not necessarily responsible for the successful outco...
Yao-Qi Guo, Hao Zhou, Jia-Yi Zhang et al.· 0 citations
LLM-based agentic systems are emerging as a new software paradigm. Modern agents are typically composed of backbone LLMs and a surrounding harness that serves as the operational software infrastructure for agent execution. As agent harnesses grow increasingly complex, agents suffer from diverse harness implementation b...
Yi-Fan Xiong, Jing-Yi Ge, Zhen-Peng Chen et al.· 0 citations
This work introduces Mubric, a mutation testing-guided approach to rubric generation, which achieves the highest overall evaluation accuracy, outperforming the strongest baseline by 7.48 percentage points.
Jiayuxuan Yang, Jie M. Zhang, Yiling Lou et al.· 0 citations
A reuse paradox is revealed: although skills are intended to be easily imported and reused, developers spend a lot of effort rewriting what the skills do, fixing skill discoverability, and translating them for different tools and languages, indicating a need for better abstractions, standardized interfaces, and automat...
Xinjian Wu, Jingzhi Gong, Gunel Jahangirova et al.· 0 citations
This paper proposes SafeKeep, an inference-time safeguard that decouples safety judgment from tool execution: it assesses requests using flattened textual tool specifications while retaining the original schema-formatted specifications for execution.
