一句话：为 LLM 修复智能体引入情节记忆与语义记忆，在同一代码库的连续 issue 流中复用经验，以更低成本大幅提升修复成功率。
仓库级程序修复中，LLM 通常对每个 issue 重置上下文，而真实维护往往是同一代码库上连续不断的工单流。作者在 SWE-Bench Pro 的分层子集（8 个项目 × 25 个实例）上研究这种序列化场景，并在每个项目内按提交时间顺序执行实例，使后续修复能利用先前轨迹。为此提出 TwinMem-Agent：将 test-patch-validate 循环与情节记忆（有据可依的演示）和语义记忆（LLM 蒸馏的笔记）配对；每个 issue 结束后记忆模块压缩轨迹、在固定预算下合并更新，并在推理时检索范例与指南，全程不微调模型权重。以 DeepSeek-V3.2 为共享骨干、使用相同 Docker 化实例，与 mini-swe-agent 对比 Pass@1 与平均推理成本，并在 25 个 Ansible 实例上对记忆模块做消融（Pass@1 与 RSR）。结果显示 TwinMem-Agent 达到 89.1% Pass@1（基线 60.0%），成本约低 18 倍（$0.0044 对 $0.0798）；移除记忆模块使 Pass@1 下降 8 个百分点、RSR 下降 12 个百分点，RSR 的更大跌幅表明跨 issue 积累的经验最直接地惠及缺陷复现，并进而提升补丁成功率。
Large language models for repository-level repair usually reset context for every issue, even though real maintenance often unfolds as a stream of tickets within the same codebase. We study this sequential regime on a stratified 8 × 25 subset of SWE-Bench Pro (eight projects, twenty-five instances each), running instances in commit-time order within each project so later repairs can draw on earlier trajectories. We present TwinMem-Agent, which pairs a test-patch-validate loop with episodic memory (grounded demonstrations) and semantic memory (LLM-distilled notes). After each issue, a memory module compacts the trajectory, merges updates under a fixed budget, and retrieves exemplars and guidelines at inference time, all without fine-tuning model weights. Using DeepSeek-V3.2 as the shared backbone and identical Dockerized instances, we compare TwinMem-Agent to mini-swe-agent on Pass@1 and average inference cost, and ablate the memory module on 25 Ansible instances measuring Pass@1 and RSR. TwinMem-Agent achieves 89.1% Pass@1 versus 60.0% for the baseline at roughly 18× lower cost ($0.0044 vs. $0.0798). Removing the memory module reduces Pass@1 by 8 pp and RSR by 12 pp, with the larger RSR drop confirming that accumulated cross-issue experience most directly benefits bug reproduction and cascades into higher patch success.