---
id: 20261002-T0-06
title: "MILO：多智能体协同进化实现自动化测试框架发现"
title_en: "MILO: Automated Harness Discovery via Multi-Agent Evolution"
url: https://ai.daily.yangsir.net/daily/20261002-T0-06
issue_date: 2026-10-02
publish_date: 2026-10-01T04:00:00.000Z
category: research
source_name: "arXiv cs.LG (ML)"
source_url: https://arxiv.org/abs/2609.38349
---

# MILO：多智能体协同进化实现自动化测试框架发现

arXiv发表论文提出MILO方法，通过编排多智能体进化来自动发现AI智能体的测试框架（harness）。现代智能体系统由AI模型和控制执行与环境交互的harness组成，harness设计对长周期任务性能影响很大，但其组合搜索空间需要大量人工投入。MILO试图自动化这一过程。

## English Version

**MILO: Automated Harness Discovery via Multi-Agent Evolution**

A paper on arXiv introduces MILO, a method that uses orchestrated multi-agent evolution to automatically discover harnesses for AI agent systems. Modern agentic systems combine an AI model with a harness that controls execution and environmental interactions. Harness design strongly affects long-horizon performance, but its combinatorial search space demands substantial human effort. MILO aims to automate this discovery process.

---

**来源**：[arXiv cs.LG (ML)](https://arxiv.org/abs/2609.38349)

**详情页**：https://ai.daily.yangsir.net/daily/20261002-T0-06

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*