发表机构
Electronics and Telecommunications Research Institute; University of Science and Technology（韩国电子通信研究院; 科学技术联合大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
YouRA提出一种持久状态架构，通过验证状态、独立控制和有状态反思实现证据可追踪的自主研究，在MLR-Bench上优于现有方法。
AI 中文摘要
端到端研究智能体现在能够生成完整的科学论文，但论文中的声明往往与已执行的实验存在偏差。这种差距是结构性的：研究状态、失败历史和声明-证据对齐在长周期流程中并未作为持久、可验证的状态加以维护。我们提出了YouRA（Your Research Agent），一种用于有状态、证据可追踪的自主研究的架构。YouRA通过整合三个组件，在研究轨迹中保留研究状态、执行证据和失败历史：一个验证状态架构（VSA），用于追踪假设、门控和证据指针；一个独立控制器，将状态和反思记录转化为生命周期、恢复和辩论/审查控制，同时将控制与执行分离；以及有状态反思，将失败记录为结构化教训，并通过有界修复、重新设计或重置来引导恢复。在MLR-Bench预定义的十任务端到端子集上，YouRA在所有三个匹配的骨干网络上均优于MLR-Agent和AI Scientist V2的标量Overall指标。一项使用MLR-Bench幻觉分类法的自动化诊断报告了四种基于事实的失败类型的交集/并集计数，数据来源诊断显示更多基于真实数据的输出。对四个组件（VSA、独立控制器、MCP工具访问和反思引导的恢复）中的每一个进行消融，支持了它们各自独立的贡献。移除任一核心状态组件都会使YouRA的性能降至完整系统之下。代码：此https URL。
英文摘要
End-to-end research agents can now produce complete scientific papers, yet manuscript claims often diverge from executed experiments. This gap is structural: research state, failure histories, and claim-evidence alignment are not maintained as persistent, verifiable state across long-horizon pipelines. We present YouRA (Your Research Agent), an architecture for stateful, evidence-traceable autonomous research. YouRA preserves research state, execution evidence, and failure history across the research trajectory by integrating three components: a Verification State Architecture (VSA) that tracks hypotheses, gates, and evidence pointers; an Independent Controller that turns state and reflection records into lifecycle, recovery, and debate/review control while separating control from execution; and Stateful Reflection that logs failures as structured lessons and routes recovery through bounded repair, redesign, or reset. On MLR-Bench's predefined ten-task end-to-end subset, YouRA improves over both MLR-Agent and AI Scientist V2 on scalar Overall across all three matched backbones. An automated diagnostic using MLR-Bench's hallucination taxonomy reports intersection/union counts for four fact-based failure types, and data-provenance diagnostic shows more real-data-based outputs. Ablating each of the four components (the VSA, the Independent Controller, MCP tool access, and reflection-guided recovery) supports their separable contributions. Removing either core-state component drops YouRA below the full system. Code: https://github.com/PrayPrey/Your-Research-Agent.
CommentsAccepted to the AACL-IJCNLP 2026 Main Conference. 21 pages, 5 figures, 17 tables
