一句话：提出 LAC 框架让多智能体在连续空间直接共享 KV-cache 通信，并实现为双链自动程序修复系统 LaMAR。
多智能体程序修复通常以文本 token 通信：每个智能体先把内部状态解码为离散符号，再由下一个智能体重新编码，这一过程既损失信息，又带来随智能体数量平方增长的 token 开销。本文提出 Latent Agent Communication（LAC），一个有理论基础的框架，形式化智能体如何在连续空间中直接共享 KV-cache 表示，并证明该通信信道具有严格更高的信息容量，同时把 token 复杂度从 O(N²) 降到 O(N)。作者将 LAC 实例化为 LaMAR 双链修复系统：Bug Analysis Chain 由四个潜在智能体协作诊断缺陷，Repair Chain 通过多智能体潜在协作精化补丁并进行跨链上下文传递，还支持对复杂缺陷的可选多轮迭代。在覆盖 Java、Python、C++ 的三个基准和三个模型族上评估，LaMAR 搭配 Qwen3-8B 在 Defects4J 上达到 24.5% 修复成功率，在 DebugBench 与 QuixBugs 上达到 62.1%–87.5%，在绝大多数设置中优于所有基线；消融实验表明缺陷分析链与潜在通信是最关键的组件。
Multi-agent systems for automated program repair communicate through text tokens, requiring each agent to decode its internal state into discrete symbols and the next agent to re-encode them. This conversion loses information and incurs token overhead that grows quadratically with the agent count. We develop Latent Agent Communication (LAC), a theoretically grounded framework that formalizes how agents can share KV-cache representations directly in continuous space, and prove it provides a communication channel with strictly higher information capacity while reducing token complexity from O(N^2) to O(N). We instantiate LAC as LaMAR, a dual-chain repair system: Bug Analysis Chain diagnoses the bug through four collaborating latent agents; Repair Chain refines the patch through multi-agent latent collaboration with cross-chain context transfer, and supports optional multi-round iteration for complex defects. Evaluated across three benchmarks spanning Java, Python, and C++ with three model families, LaMAR with Qwen3-8B achieves 24.5% repair success rate on Defects4J and 62.1%--87.5% on DebugBench and QuixBugs, outperforming all baselines in the large majority of settings. Ablation experiments identify the bug analysis chain and latent communication as the most critical components.
@inproceedings{paperbot12474,
title = {LaMAR: Latent Multi-agent Collaboration via KV-Cache Communication for Automated Program Repair},
author = {Rui Jiao and Yue Zhang and Zhexuan Feng and Jinku Li},
booktitle = {ASE},
year = {2026},
doi = {10.1145/3832783.3837423}
}