Reusable skills are a key mechanism for extending agent capabilities, allowing agents to accumulate experience and solve increasingly complex tasks. Yet most existing skill-learning methods store reusable experience as text-only assets, such as instructions, reasoning traces, or summarized trajectories. We argue that this text-only paradigm creates a fundamental bottleneck for visual-centric tasks, where reusable knowledge often depends on spatial layout, visual grounding, fine-grained appearance, and localized state changes. To address this limitation, we propose \textbf{\NAME}, a multimodal skill paradigm that combines declarative textual logic with explicit visual support. We distinguish three reusable forms: static priors for stable spatial conventions, dynamic priors for in-situ visual working memory, and interleaved visual skills that bind ordered text steps to the source frames, screenshots, or page regions that justify them. Rather than only describing what to do, visual skills also encode where to look, how to inspect, and how to verify visual outcomes. To scale visual-skill construction, we introduce \textbf{\SYSTEM}, an automatic system that converts agent experience into reusable multimodal skills by preserving textual reasoning, spatial references, visual boundaries, and interaction patterns from task trajectories. Experiments on GUI and other visual-centric tasks show that visual skills consistently outperform text-only skills, particularly when success requires spatial correspondence, visual evidence, and state-aware interaction. These results support our central position: reusable agent skills should go beyond text and become multimodal assets for future multimodal agents.
AtlasVA: Self-Evolving Visual Skill Memory for Teacher-Free VLM Agents
Ant Group(蚂蚁集团);University of Science and Technology of China(中国科学技术大学);Westlake University(西湖大学);University of Michigan - Ann Arbor(密歇根大学-安娜堡分校);Sun Yat-sen University(中山大学)
University of Zagreb(萨格勒布大学);University of Twente(特文特大学);University of Groningen(格罗宁根大学);Radboud University Medical Center(拉德堡德大学医学中心);Canisius-Wilhelmina Hospital(卡尼修斯-威廉明娜医院)
Interactive Semantic Map Representation for Skill-based Visual Object Navigation
Moscow Institute of Physics and Technology(莫斯科物理技术学院);Artificial Intelligence Research Institute(人工智能研究院);Federal Research Center for Computer Science and Control of Russian Academy of Sciences(俄罗斯科学院联邦计算机科学与控制研究中心)
Looking Beyond the Scale: Do Surgical Skill Models Learn Transferable Representations Across Assessment Rubrics?
National Center for Tumor Diseases (NCT)(国家肿瘤疾病中心);DKFZ(德国癌症研究中心);Faculty of Medicine and University Hospital Carl Gustav Carus(卡尔·古斯塔夫·卡鲁斯医学院及大学医院);TUD Dresden University of Technology(德累斯顿工业大学);Helmholtz-Zentrum Dresden-Rossendorf (HZDR)(德累斯顿-罗森多夫亥姆霍兹中心);BMFTR Research Hub 6G-Life(BMFTR 6G生活研究中心);Deutsches Krebsforschungszentrum (DKFZ)(德国癌症研究中心);The Centre for Tactile Internet with Human-in-the-Loop (CeTI)(人在回路触觉互联网中心)
Seeing Is Not Screening: Multimodal Hidden Instruction Attacks on Agent Skill Scanners
Nanyang Technological University, Singapore(南洋理工大学,新加坡);Chongqing University, China(重庆大学);Northeastern University, China(东北大学);University of Chinese Academy of Sciences, China(中国科学院大学);Beihang University, China(北航大学)
Learning Skills from Historical Action Trajectories: Action Experience Dictionary for World Action Models
Shenyang Institute of Automation, Chinese Academy of Sciences(中国科学院沈阳自动化研究所);Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学);Anhui University(安徽大学);Xiaomi Corporation(小米公司);Fudan University(复旦大学);University of Trento(特伦托大学)
Ag2Manip: Learning Novel Manipulation Skills with Agent-Agnostic Visual and Action Representations
Beijing Institute for General Artificial Intelligence (BIGAI)(北京通用人工智能研究院);Tsinghua University(清华大学);Peking University(北京大学);University of California, Los Angeles(加利福尼亚大学洛杉矶分校)