---
格式版本: 2
标题: "Eureka: Task-Conditioned Meta-Agent Orchestration for Scientific Discovery"
原文链接: "https://arxiv.org/abs/2608.19047"
发布日期: "2026-08-19"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Wed, 19 Aug 2026 15:40:21 UTC (1,804 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T14:37:09+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T14:35:51+08:00"
入库时间: "2026-08-20T06:37:09.335Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=Meta&searchtype=all"
匹配关键词:
  - "AI"
相关厂家:
  - "Meta"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 10
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "该arXiv论文为AI智能体编排研究，主题为科学发现与数学证明，与超节点、AI Rack、机柜级AI基础设施完全无关，无任何相关技术、商业或部署信息。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T14:38:15+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 5
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 0
AI摘要: "研究者提出任务条件化Meta-Agent架构Eureka，将长时程任务编译为动态义务图，通过宏代理与成本收益门控演化使架构匹配任务认知结构。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:17:25.513Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.19047"
---

## Computer Science > Artificial Intelligence

## Title:Eureka: Task-Conditioned Meta-Agent Orchestration for Scientific Discovery

Authors:[Alizer Wong](https://arxiv.org/search/cs?searchtype=author&query=Wong,+A), [Heng Cui](https://arxiv.org/search/cs?searchtype=author&query=Cui,+H), [Yi Tan](https://arxiv.org/search/cs?searchtype=author&query=Tan,+Y), [Xiongchao Zhan](https://arxiv.org/search/cs?searchtype=author&query=Zhan,+X), [Liang Lin](https://arxiv.org/search/cs?searchtype=author&query=Lin,+L), [Yuxiang Guo](https://arxiv.org/search/cs?searchtype=author&query=Guo,+Y), [Zhaorong Dai](https://arxiv.org/search/cs?searchtype=author&query=Dai,+Z), [Zixin Zeng](https://arxiv.org/search/cs?searchtype=author&query=Zeng,+Z), [Wenyuan Li](https://arxiv.org/search/cs?searchtype=author&query=Li,+W)

[View PDF](https://arxiv.org/pdf/2608.19047) [HTML (experimental)](https://arxiv.org/html/2608.19047v1)

> Abstract:We present Eureka, a task-conditioned Meta-Agent architecture that compiles long-horizon tasks into dynamic obligation graphs with explicit acceptance semantics. During execution, Eureka forms Macro-Agents with specialized state, memory, operators, tools, verifiers, and local topology via receding-horizon planning, architecture promotion, and minimal-sufficient compilation. When bottlenecks recur, cost-benefit-gated evolution updates the local architecture under constraints. Theoretically, we establish results on regret, planning invalidation, amortization, subtree interfaces, serializability, and verification. Experimentally, Eureka completes 170/170 recursive tasks and generates 3,948 certificates with no false acceptances. Active context compresses median input from 9,490 to 4,005 tokens; incremental processing avoids 65.38% recomputation across 12,000 tasks; 16,000 concurrent executions serialize consistently. The same Meta-Agent instantiates a Theory-Discovery Agent and a Math/Conjecture Agent. The former yields structural results in quantum-process and spacetime theory. The latter identifies bottlenecks in Riemann Hypothesis research and advances a positivity certificate for Suzuki's localized Weil quadratic form to 0 < a <= 69/200 = 0.345, reaching ~99.55% of (log 2)/2. These results suggest that scientific-agent capability depends not only on the base model but on whether an architecture can be formed to match the task's cognitive structure.

| Comments: |  |
| --- | --- |
| Subjects: | Artificial Intelligence (cs.AI); Number Theory (math.NT) |
| Cite as: | [arXiv:2608.19047](https://arxiv.org/abs/2608.19047) \[cs.AI\] |
|  | (or [arXiv:2608.19047v1](https://arxiv.org/abs/2608.19047v1) \[cs.AI\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.19047](https://doi.org/10.48550/arXiv.2608.19047) |

## Submission history

From: Zixin Zeng \[[view email](https://arxiv.org/show-email/13cbe64a/2608.19047)\]  
**\[v1\]** Wed, 19 Aug 2026 15:40:21 UTC (1,804 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.19047) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))