[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83073-en":3,"doc-seo-83073-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},83073,1374391975076,"Riley","https://ap-avatar.wpscdn.com/avatar/14000253ca4ec9f6853?x-image-process=image/resize,m_fixed,w_180,h_180&k=1783305029341752051",8,"Research & Report","Task Decomposition-Guided Reranking for Adaptive Agent Skill Retrieval","Skill usage enhances modern agent systems in completing complex tasks, yet scaling skill libraries makes accurate skill selection increasingly difficult. Ambiguous semantic matching often occurs between a single task requirement and multiple generic but semantically similar candidate skills. Existing approaches also ignore dynamic effects from task difficulty and skill applicability during selection. SkillReranker is introduced as an inference-time reranking framework that decomposes tasks and skills, models structured task–skill correspondences via a directed acyclic execution graph, and adaptively scores candidates per interval, improving performance on ALFWorld and ScienceWorld while reducing steps and token usage.","Task Decomposition-Guided Reranking for Adaptive Agent Skill Retrieval  \nYanping Chen 1 , Weijie Shi2 , Wen Yang 1 , Jiajie Xu 1  \n1 School of Computer Science and Technology, Soochow University  \n2The Hong Kong University of Science and Technology  \narXiv :2607 .06283v 1 [ cs .AI ] 7 Jul 2026  \nAbstract  \nSkill usage can significantly enhance the ability of modern agent systems to complete complex tasks. However, the growing scale of skill libraries makes accurate skill selection increasingly challenging. In real-world scenarios, ambiguous semantic matching often arises between a specific task requirement and multiple generic yet semantically similar candidate skills. Moreover, existing methods tend to overlook the dynamic influence of task difficulty and skill applicability when selecting the optimal target skill set. To address these issues, we propose SkillReranker, an inference-time reranking framework for adaptive skill selection. Specifically, we first perform semantic decomposition on both the task and skill sides, yielding informative subtask and execution-state descriptions as well as transition-state descriptions that characterize each skill’s functionality. These descriptions are then used to construct a directed acyclic execution graph, where intermediate task states are modeled as nodes and candidate skills as edges, thereby establishing a structured task–skill correspondence. On this basis, SkillReranker determines whether each state node satisfies the split condition to identify subtask intervals. For each task interval, we employ a cross-encoder to perform comprehensive scoring over candidate skills and select the most suitable ones to form the final target skillset. Experiments on ALFWorld and ScienceWorld with three backbone LLMs show that SkillReranker effectively improves task performance, reduces environment interaction steps, and lowers token consumption compared with existing skill selection baselines.  \n1 Introduction  \nSkills have emerged as an effective mechanism for enhancing the ability of large language model (LLM) agents to solve complex tasks (Liu et al. 2025; Wang et al. 2026b; Li et al. 2026b) . Rather than requiring agents to rely entirely on the implicit reasoning and action generation of the underlying LLM, skill libraries explicitly organize both domain knowledge and procedural guidance, thereby providing more stable execution support (Xu and Yan 2026; Jiang et al. 2026) . However, as skill libraries continue to grow in scale (Li et al. 2026a), the functional boundaries among skills become increasingly blurred. Accurately selecting the skills that best match the requirements of a given task from a large candidate pool remains a key challenge for LLM-based agents (Su et al. 2026; Cho, Kang, and Kim 2026) .  \nTask:  \nWelcome to TextWorld, ALFRED! Your task is to: put a hot cup in cabinet.  \nCandidate skill A: alfworld-heat-object-with-appliance  \n_  \nDescription: Uses a heating appliance (microwave, stoveburner, oven) to apply heat to a specified object…  \nCandidate skill  \nB: alfworld-object-heater  \nSimilar?  \n_  \nDescription: Heats a specified object using an available heating appliance (e.g., microwave, stoveburner)…  \nDistinction:  \nSkill A has no prerequisites.  \nSkill B has a prerequisite: The target object must be in your inventory.  \nSkill A: detailed workflow (Navigate & Locate → Acquire Object → Navigate to Appliance → Prepare Appliance → Execute Heating)  \nSkill B: concise workflow (Navigate → Check state → Heat)  \nFigure 1: Task-matching difficulty caused by semantic similarity between skills. Thetaskis selected from the ALFWorld dataset.  \nThis challenge first arises from a granularity gap between task requirements and skill descriptions. In real-world scenarios, task requirements are typically concrete and concise, whereas skill descriptions tend to be more general, often targeting a class of similar tasks or a reusable operational pattern (Liang et al. 2026a; Zheng et al. 202","cbCairNcZHM0IH4F","https://ap.wps.com/l/cbCairNcZHM0IH4F","pdf",2096595,3,1,10,"English","en",105,"# Introduction\n## Skill selection challenges\n## Proposed SkillReranker approach\n## Task–skill decomposition and execution graph\n## Cross-encoder scoring for adaptive skill sets","[{\"question\":\"What problem does SkillReranker address in adaptive agent skill retrieval?\",\"answer\":\"It addresses incorrect skill selection caused by semantic ambiguity among functionally different skills and by neglecting the dynamic impact of task difficulty and skill applicability when choosing a target skill set.\"},{\"question\":\"How does SkillReranker use task and skill decomposition?\",\"answer\":\"It decomposes task descriptions into subtasks, intermediate states, and transition-state descriptions, while extracting state descriptions from skills to characterize applicable conditions and expected effects.\"},{\"question\":\"How are candidate skills ranked and selected in SkillReranker?\",\"answer\":\"SkillReranker builds a directed acyclic execution graph mapping intermediate task states to skills, splits the task into intervals using a split condition on state nodes, and uses a cross-encoder to score candidate skills per interval to form the final target skill set.\"}]",1784185021,25,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"task-decomposition-guided-reranking-for-adaptive-agent-skill-retrieval","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":20},"https://docshare.wps.com/document/research-report/",{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/task-decomposition-guided-reranking-for-adaptive-agent-skill-retrieval/83073/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-25","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does SkillReranker address in adaptive agent skill retrieval?","Question",{"text":75,"@type":76},"It addresses incorrect skill selection caused by semantic ambiguity among functionally different skills and by neglecting the dynamic impact of task difficulty and skill applicability when choosing a target skill set.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does SkillReranker use task and skill decomposition?",{"text":80,"@type":76},"It decomposes task descriptions into subtasks, intermediate states, and transition-state descriptions, while extracting state descriptions from skills to characterize applicable conditions and expected effects.",{"name":82,"@type":73,"acceptedAnswer":83},"How are candidate skills ranked and selected in SkillReranker?",{"text":84,"@type":76},"SkillReranker builds a directed acyclic execution graph mapping intermediate task states to skills, splits the task into intervals using a split condition on state nodes, and uses a cross-encoder to score candidate skills per interval to form the final target skill set.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":22,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":22,"slug":133},"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]