Agensh: Scaling Organizational Intelligence to 1,024 Agents
- 类型：arxiv
- 标识：2609.26781
- 链接：https://arxiv.org/abs/2609.26781
- 主分类：agent
- 形态：method
- TLDR：A multi-agent system can reduce latency on complex tasks by executing work concurrently. Several pioneering harness frameworks support multi-agent systems. However, the scalability of current multi-agent harnesses is often constrained by a central orchestrator's capacity to allocate tasks and coordinate workers. To address this limitation, we introduce Agensh, a scalable self-organized multi-agent harness without a central orchestrator: concurrent workers execute a multi-agent cooperation loop, continuously gathering context, claiming and self-assigning sub-tasks, taking action and sharing fin
- 副分类：evaluation
- 待LLM分类：否
- 标题中文：Agensh：将组织智能扩展到 1,024 个智能体
- TLDR中文：多智能体系统可通过并发执行任务来降低复杂任务的时延。多个开创性 harness 框架支持多智能体系统。然而，现有 multi-agent harness 的可扩展性常受限于中心编排器分配任务与协调 worker 的能力。为解决该局限，我们提出 Agensh，一种无需中心编排器的可扩展自组织 multi-agent harness：并发 worker 运行多智能体协作循环，持续收集上下文、自主认领并自分配子任务、执行动作并共享完成
- 来源文件：
- /inbox/tom/_candidates/2026-09-23-agent-rag-longcontext-candidates.json