[2604.03632] Persistent Cross-Attempt State Optimization for Repository-Level Code Generation

Computer Science > Software Engineering arXiv:2604.03632 (cs) [Submitted on 4 Apr 2026] Title:Persistent Cross-Attempt State Optimization for Repository-Level Code Generation Authors:Ruwei Pan, Jiangshuai Wang, Qisheng Zhang, Yueheng Zhu, Linhao Wu, Zixiong Yang, Yakun Zhang, Lu Zhang, Hongyu Zhang View a PDF of the paper titled Persistent Cross-Attempt State Optimization for Repository-Level Code Generation, by Ruwei Pan and 8 other authors View PDF HTML (experimental) Abstract:Large language models (LLMs) have achieved substantial progress in repository-level code generation. However, solving the same repository-level task often requires multiple attempts, while existing methods still optimize each attempt in isolation and do not preserve or reuse task-specific state across attempts. In this paper, we propose LiveCoder, a novel framework for repository-level code generation based on cross-attempt knowledge optimization. LiveCoder maintains persistent task-specific state from prior attempts to guide subsequent generation. This state includes success knowledge, which captures reusable signals from previously strong repositories, failure knowledge, which records unsuccessful outcomes and their diagnostic signals, and a historical-best repository, which preserves the strongest result found so far and prevents regression. These components collectively transform repeated repository generation into a persistent, knowledge-driven optimization process. We evaluate LiveCoder using...