一句话:通过持久记忆与检索-行动-蒸馏循环,提升仓库级问题解决的通过率与效率。
本文面向长周期、仓库级软件工程工作流中智能体难以复用历史经验的问题,提出 Exp-SWE-Agent。该方法在 Mini-SWE-Agent 框架上增加轻量级持久记忆模块,通过持续的“检索-行动-蒸馏”循环,从过往问题解决轨迹中抽取结构化的事实性与经验性知识,并在后续任务中作为策略提示使用。作者在 SWE-Bench Pro 的顺序 8×25 子集上评估,问题解决率(Pass%)从 42.7% 提升至 47.5%,同时持续减少交互步数。结果表明,结构化持久记忆能有效提升智能体在长周期仓库级软件工程任务中的准确性与效率。
We present Exp-SWE-Agent, our submission to the FSE-AIWare 2026 MemComp challenge. Exp-SWE-Agent augments the Mini-SWE-Agent framework with a lightweight persistent memory module. Through a continuous retrieve-act-distill cycle, the agent extracts structured factual and experiential knowledge from past issue-resolution trajectories and uses them as strategic hints for subsequent tasks. Evaluated on a sequential 8 × 25 subset of SWE-Bench Pro, Exp-SWE-Agent increases the issue resolution rate (Pass%) from 42.7% to 47.5% while consistently reducing the number of interaction steps. These results demonstrate that structured persistent memory effectively enhances both the accuracy and efficiency of agents in long-horizon, repository-level software engineering workflows.