Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.05004 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914306645295104 |
|---|---|
| author | Lin, Zexin Yu, Jiachen Zhang, Haoyang Li, Yuzhao Li, Zhonghang Yang, Yujiu Wang, Junjie Ji, Xiaoqiang |
| author_facet | Lin, Zexin Yu, Jiachen Zhang, Haoyang Li, Yuzhao Li, Zhonghang Yang, Yujiu Wang, Junjie Ji, Xiaoqiang |
| contents | Large language models are enabling language-conditioned agents in interactive environments, but highly cooperative tasks often impose two simultaneous constraints: sub-second real-time coordination and sustained multi-episode adaptation under a strict online token budget. Existing approaches either rely on frequent in-episode reasoning that induces latency and timing jitter, or deliver post-episode improvements through unstructured text that is difficult to compile into reliable low-cost execution. We propose CoWork-X, an active co-evolution framework that casts peer collaboration as a closed-loop optimization problem across episodes, inspired by fast--slow memory separation. CoWork-X instantiates a Skill-Agent that executes via HTN (hierarchical task network)-based skill retrieval from a structured, interpretable, and compositional skill library, and a post-episode Co-Optimizer that performs patch-style skill consolidation with explicit budget constraints and drift regularization. Experiments in challenging Overcooked-AI-like realtime collaboration benchmarks demonstrate that CoWork-X achieves stable, cumulative performance gains while steadily reducing online latency and token usage. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_05004 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CoWork-X: Experience-Optimized Co-Evolution for Multi-Agent Collaboration System Lin, Zexin Yu, Jiachen Zhang, Haoyang Li, Yuzhao Li, Zhonghang Yang, Yujiu Wang, Junjie Ji, Xiaoqiang Computation and Language Artificial Intelligence Large language models are enabling language-conditioned agents in interactive environments, but highly cooperative tasks often impose two simultaneous constraints: sub-second real-time coordination and sustained multi-episode adaptation under a strict online token budget. Existing approaches either rely on frequent in-episode reasoning that induces latency and timing jitter, or deliver post-episode improvements through unstructured text that is difficult to compile into reliable low-cost execution. We propose CoWork-X, an active co-evolution framework that casts peer collaboration as a closed-loop optimization problem across episodes, inspired by fast--slow memory separation. CoWork-X instantiates a Skill-Agent that executes via HTN (hierarchical task network)-based skill retrieval from a structured, interpretable, and compositional skill library, and a post-episode Co-Optimizer that performs patch-style skill consolidation with explicit budget constraints and drift regularization. Experiments in challenging Overcooked-AI-like realtime collaboration benchmarks demonstrate that CoWork-X achieves stable, cumulative performance gains while steadily reducing online latency and token usage. |
| title | CoWork-X: Experience-Optimized Co-Evolution for Multi-Agent Collaboration System |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2602.05004 |