Survey on Agentic Software Issue Resolution with LLMs

A new survey paper on arXiv (2512.22256) reviews the emerging field of agentic software issue resolution using large language models (LLMs). The paper highlights that resolving real-world software issues is complex, requiring long-horizon reasoning, iterative exploration, and feedback-driven decision-making, which go beyond conventional single-step approaches. The authors note that LLM-based agentic systems have become a promising research direction, with rapid growth in literature. The survey aims to consolidate advances in this area, potentially improving software maintenance efficiency and quality. The paper is categorized as a replace-cross announcement type and was published on arXiv.