跳到正文
arXiv cs.AI· Yunxiao Shi (Australian National University), Tingwei Chen (University of Technology Sydney), Dannong Xu (Australian National University), Ahad N. Zehmakan (Australian National University), Jialie Shen (City St George's, University of London), Zhendong Chu (Squirrel Ai Learning), Qingsong Wen (Squirrel Ai Learning)·· 4 小时前AI 评分38

智能体需要什么?Bundle-Aware 智能体资源组合方法

What Does an Agent Need? Bundle-Aware Agent Resource Composition

AI 导读

研究者提出 AgentComposition 任务与 Bundle-Aware 资源组合方法,联合选择主干模型与工具技能包以最大化端到端执行效用。该方法在 67.73% 的 AgentSelect 查询中于十个候选内恢复完整参考配置,约为最强基线的两倍,并在 SkillsBench 上领先所有基线。

正文 · 原文

Authors:Yunxiao Shi (1), Tingwei Chen (2), Dannong Xu (1), Ahad N. Zehmakan (1), Jialie Shen (3), Zhendong Chu (4), Qingsong Wen (4) ((1) Australian National University, (2) University of Technology Sydney, (3) City St George's, University of London, (4) Squirrel Ai Learning)

View PDF HTML (experimental)

Abstract:An agent resource's value depends on its backbone and companion resources and is known only after execution. Token-based generators remain tied to their training catalog, while independent resource scoring misses complementarities and reused bundles target other tasks. We formulate Task-Conditioned Agent Resource Composition (AgentComposition): jointly selecting a backbone model and a tool-skill bundle to maximize end-to-end execution utility. Our Retrieval-Grounded Bundle-Aware Agent Generation retrieves candidate resources and past configurations of similar tasks. A fine-tuned generator reads resource descriptions and recombines candidates through constrained beam search, allowing new resources without retraining. A bundle-aware critic ranks complete configurations using a large language model (LLM) teacher and historically successful configurations as execution proxies, and adapts to execution feedback when available. We evaluate reference recovery, offline execution with judged and native rewards, and online recommendation with real execution, building AgentSelect-Exec with 5,060 executed configurations. Our method recovers the complete reference configuration within ten candidates for 67.73% of AgentSelect queries, nearly twice the strongest baseline, and leads all baselines on SkillsBench. The proxy-trained critic raises the top-ranked agent's execution score from 7.58 to 8.37 out of 10; execution feedback increases it to 8.78, above the strongest baselines. Four target-domain queries suffice to adapt the generator to SkillsBench. Agents recommended online outperform the strongest baselines under both an LLM judge and native verifiers. These results support learning and transferring joint resource composition from task experience.
Comments: 26 pages, including appendices. Code: this https URL
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.22030 [cs.AI]
  (or arXiv:2611.22030v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.22030

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Yunxiao Shi [view email]
[v1] Tue, 17 Nov 2026 00:45:03 UTC (655 KB)

来源:arXiv cs.AI · arxiv.org