让科研自动化系统学会从试错中自我进化，不再只生成论文。
Sibyl-AutoResearch: Autonomous Research Needs Self-Evolving Trial-and-Error Harnesses, Not Paper Generators
- 构建可自我进化的试错框架，将实验结果转化为后续研究动作。
- 实测发现8个关键转化事件，平均1轮内完成反馈闭环，最长3轮。
- 适合关注自主科研系统可追溯性与鲁棒性的研究者。
当前自主科研系统虽能执行科学工作流，但缺乏研究判断力。问题在于：弱证据变成陈述、初步信号演变为广泛结论、记忆仅以文本形式留存、重复失败未影响后续行为。为此，提出Sibyl-AutoResearch框架，核心是科学试错引擎（Scientific Trial-and-Error Harnesses）。该引擎支持有限范围实验，保留正负结果，并将经验导入后续规划、验证、论断范围、调度、批判、写作及引擎修复。通过两个可审计的转换单元实现：试错到行为转换（链接实验信号与后续行动），试错到引擎行为转换（链接重复失败与系统更新）。在基于文件的SIBYL系统中实现，可追踪状态、角色、记忆、门控与产物痕迹。回溯审计识别出8个高置信度转化事件，中位延迟为1次迭代，最大延迟3次；故障恢复注册表显示5类自然发生的问题（如重复结果、过时数据、不支持统计）被阻断、降级或引导至修复。这些痕迹未做性能对比，但证明转换机制可在真实自治研究空间中可复现。代码与系统已开源。
原文摘要 · Abstract (English)
Autonomous research systems increasingly make the scientific workflow executable: agents can propose ideas, run code, inspect results, and draft papers. But executable workflows do not by themselves produce research judgment. We analyze where current systems lose trial experience: weak evidence becomes prose, pilot signals become broad claims, memory remains textual, and recurring process failures do not change later behavior. We introduce Sibyl-AutoResearch, a self-evolving AutoResearch framework built around Scientific Trial-and-Error Harnesses. A harness lets agents run bounded trials, preserve positive and negative outcomes, and route lessons into later planning, validation, claim scope, scheduling, critique, writing, and harness repair. We formalize this through two auditable conversion units: trial-to-behavior conversion, which links trial signals to later research actions, and trial-to-harness-behavior conversion, which links recurring process failures to system updates. We implement the framework in SIBYL, a file-backed autonomous research system that exposes the state, roles, memory, gates, and artifact traces needed to inspect these conversion paths. A retrospective audit identifies eight high-confidence conversion events, with a median latency of one iteration and a maximum latency of three iterations. A recovered-failure registry further shows how five naturally occurring failure classes, including duplicate results, stale numbers, and unsupported statistics, were blocked, downgraded, or routed into later repair. These traces do not establish a comparative performance claim; they show that the proposed conversion units are recoverable from realistic autonomous-research workspaces. The SIBYL framework and system are available at https://github.com/Sibyl-Research-Team/AutoResearch-SibylSystem.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。