The paper introduces EvoSkill-GUI, a training‑free framework that enables GUI agents to evolve their skills during deployment. Each skill is packaged with metadata, executable plans, and recovery rules, and the system follows a reflect‑revise‑reuse loop where the agent instantly revises skills based on execution feedback. Experiments on MobileWorld, AndroidWorld, and OSWorld show consistent performance gains up to +16.2% without any additional training.
By Bofan Chen, Boxuan Zhang, Fei Tang, Zhengxi Lu, Yong Du, Tongbo Chen, Weiming Lu, Jun Xiao, Yueting Zhuang, Yongliang Shen
arXiv:2607. 26598v1 Announce Type: cross Abstract: Large language model (LLM) agents may recover from a failure within an episode or after a retry, yet the same execution failure can recur in later tasks because post-episode feedback rarely revises the persistent harness that guides future interactions.
By Yuetian Du, Yucheng Wang, He Xu, Jiexu Xu, Shanwen Tan, Bing Zhao, Boyu Yang, Zhijie Xu, Ming Kong, Hu Wei, Jie Liu, Qiang Zhu
arXiv:2604. 08377v2 Announce Type: replace Abstract: Large language model (LLM) agents such as OpenClaw rely on reusable skills to perform complex tasks, yet these skills remain largely static after deployment.
By Ziyu Ma, Shidong Yang, Yuxiang Ji, Xucong Wang, Yong Wang, Yiming Hu, Tongwen Huang, Xiangxiang Chu
arXiv:2609.13667v1 Announce Type: new
Abstract: Geospatial agents are increasingly expected to support recurring and evolving analytical tasks rather than execute isolated workflows. In such settings...
By Han Luo, Xian Xu, Yinhe Liu, Yanfei Zhong
SkillGLoW introduces a new way for large language model agents to self‑improve by consolidating procedural skills shared across related tasks. Instead of storing all skills in a single global document or a flat per‑task pool, SkillGLoW aggregates local skills into procedural families, compresses them into de‑instantiated global priors, and regenerates instance‑specific details on demand. Experiments on four diverse benchmarks show that these priors improve performance by an average of 17.2 points over a no‑skill baseline, are more compact than per‑task pools, and enable better transfer to unseen tasks.
By Ao Yan, Xin Zhang, Jiawei Du, Joey Tianyi Zhou
arXiv:2606. 14239v1 Announce Type: new Abstract: Agent skills are structured procedural packages that guide frozen LLM agents in specialized workflows.
By Haowen Gao, Haoran Chen, Can Wang, Shasha Guo, Liang Pang, Zhaoyang Liu, Huawei Shen, Xueqi Cheng
The paper introduces Agent S4, an autonomous system that learns to design and refine user interfaces for computer‑use agents by observing task execution without explicit success labels. By evolving a repository of interfaces through population‑based selection on unseen validation tasks, Agent S4 achieves state‑of‑the‑art performance on OSWorld V2 and significant gains on ScienceBoard, outperforming large commercial models at lower cost. The work demonstrates that interface design can be treated as an agent capability, enabling continuous improvement beyond fixed, hand‑crafted interfaces.
By Saaket Agashe, Yuzhe Yang, Gonzalo Gonzalez-Pumariega, Bhuvan Chandra Koduru, Chih-Lun Li, Jiachen Yang, Ang Li, Xin Eric Wang
arXiv:2608.30760v1 Announce Type: new
Abstract: Recent studies have shown that multimodal large language models (MLLMs) can serve as embodied agents, translating language instructions and visual obse...
By Ziyi Bai, Siqi Li, Tinglei Huang, B\"orje F. Karlsson
Computer-using agents can perceive rich software interfaces, yet their decisions often lack visual procedural memory: they may recognize individual controls without identifying which familiar workflow is active, which control matters next, or what evidence would confirm progress. Raw interaction traces preserve such information but are long and noisy to condition on, whereas text-only skills often omit the visual state that makes a procedure applicable.
AgentHorizon is a benchmark comprising 1,373 computer‑use tasks with instruction‑trajectory pairs collected from 166 hours of human‑recorded activity across three operating systems. The benchmark uses a paired design that swaps instructions to create negative tasks, testing judges’ ability to distinguish successful trajectories from those that completed a similar but incompatible request. Eleven judges were evaluated, with GPT‑5.5 achieving the highest balanced accuracy of 80.9% on the full trajectory subset, while tool‑use benefits some models but harms others, revealing large variability in judges’ performance.
By Xing Han L\`u, Dheeraj Vattikonda, Sina Hajimiri, Fatemeh Pesaran Zadeh, Parishad BehnamGhader, Ghazwa Darwiche, Amirhossein Kazemnejad, Christopher Pal, Alexandre Drouin, Siva Reddy
arXiv:2607. 13854v2 Announce Type: replace Abstract: Multimodal agents that think with images iteratively manipulate visual evidence and invoke tools across many steps.
By Ru Zhang, Weijie Qiu
arXiv:2605. 18401v2 Announce Type: replace-cross Abstract: Long-horizon LLM agents generate traces that could become reusable experience, but raw trajectories are noisy, local, and hard to govern.
By Hongyi Liu, Haoyan Yang, Tao Jiang, Bo Tang, Feiyu Xiong, Yuyu Luo, Zhiyu Li