ARIS：基于对抗性多智能体协作的自主研究
ARIS: Autonomous Research via Adversarial Multi-Agent Collaboration
May 4, 2026
作者: Ruofeng Yang, Yongcan Li, Shuai Li
cs.AI
摘要
本报告介绍ARIS（自动睡眠研究系统），一个用于自主研究的开源研究框架，涵盖其架构设计、保障机制及早期部署经验。基于大语言模型的智能体系统性能既取决于模型权重，也依赖于管控信息存储、检索与呈现方式的研究框架。针对长周期研究流程，核心失效模式并非显性中断，而是看似合理但缺乏依据的成功——长期运行的智能体可能产生证据不完整、误报或隐性沿袭执行者预设框架的结论。为此，我们提出ARIS研究框架，其默认配置通过跨模型对抗性协作协调机器学习研究流程：由执行模型推动研究进展，同时推荐采用不同模型家族的评审模型对中间成果进行批判性审阅并提请修订。
ARIS采用三层架构设计。执行层提供65+个可复用的Markdown定义技能、基于MCP的模型集成、支持迭代复用历史发现的持久化研究维基，以及确定性图表生成功能。编排层协调五种端到端工作流，配备可调节强度参数与可配置的评审模型路由机制。保障层包含三阶段实验结论验证流程：完整性核验、结果与结论映射、以及通过对比稿件陈述与结论分类账及原始证据的结论审计；此外还集成五轮科学编辑流水线、数学证明检查器及PDF渲染效果视觉审查。原型自改进循环会记录研究轨迹并提出框架优化建议，所有改进方案均需通过评审批准后方可采纳。
（注：根据技术文档翻译规范，MCP保持英文缩写原貌；"adversarial collaboration"译为"对抗性协作"以体现学术语境；"claim ledger"创新译为"结论分类账"以保持会计学隐喻；长难句按中文表达习惯进行合理切分与语序调整。）
English
This report describes ARIS (Auto-Research-in-sleep), an open-source research harness for autonomous research, including its architecture, assurance mechanisms, and early deployment experience. The performance of agent systems built on LLMs depends on both the model weights and the harness around them, which governs what information to store, retrieve, and present to the model. For long-horizon research workflows, the central failure mode is not a visible breakdown but a plausible unsupported success: a long-running agent can produce claims whose evidential support is incomplete, misreported, or silently inherited from the executor's framing. Therefore, we present ARIS as a research harness that coordinates machine-learning research workflows through cross-model adversarial collaboration as a default configuration: an executor model drives forward progress while a reviewer from a different model family is recommended to critique intermediate artifacts and request revisions. ARIS has three architectural layers. The execution layer provides more than 65 reusable Markdown-defined skills, model integrations via MCP, a persistent research wiki for iterative reuse of prior findings, and deterministic figure generation. The orchestration layer coordinates five end-to-end workflows with adjustable effort settings and configurable routing to reviewer models. The assurance layer includes a three-stage process for checking whether experimental claims are supported by evidence: integrity verification, result-to-claim mapping, and claim auditing that cross-checks manuscript statements against the claim ledger and raw evidence, as well as a five-pass scientific-editing pipeline, mathematical-proof checks, and visual inspection of the rendered PDF. A prototype self-improvement loop records research traces and proposes harness improvements that are adopted only after reviewer approval.