[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82914-en":3,"doc-seo-82914-105":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},82914,8796095461610,"Oliver","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","EvoAgentBench：通过能力迁移实现智能体自我进化的基准测试","Agent self-evolution in long-horizon LLM systems is largely procedural: useful experience should translate into reusable procedures for searching, debugging, and verification. Existing evaluations do not separate this procedural transfer from task solving or information retention. EvoAgentBench benchmarks self-evolution via Ability-guided transfer across web research, algorithmic reasoning, software engineering, and knowledge work, extracting trace-grounded Abilities, building domain Ability Graphs, and enforcing verified, instance-disjoint support in a 528/267 split.","EvoAgentBench: Benchmarking Agent Self-Evolution via Ability Transfer Xingze Gao1,2 Chuanrui Hu2 † Hongda Chen2,3 Pengfei Yao2 Zhao Wang2 Yi Bai2 Zhengwei Wu2 Yunyun Han2 Xiaofeng Cong3 Jie Gui3 Yafeng Deng2 * Teng Li1 *  \n1Anhui University 2EverMind, Shanda Group 3 Southeast University  \n{xingze.gao, chuanrui.hu, hongda.chen, [yaopengfei}@shanda.com](yaopengfei}@shanda.com)[ ](yaopengfei}@shanda.com){[zhao.wang](zhao.wang) , baiyi, hanyunyun, [dengyafeng}@shanda.com](dengyafeng}@shanda.com)[ ](dengyafeng}@shanda.com)[zhengwei.wu@evermind.ai](zhengwei.wu@evermind.ai) [tenglwy@gmail.com](tenglwy@gmail.com)  \n[cxf_svip@163.com](cxf_svip@163.com) [guijie@ustc.edu](guijie@ustc.edu)  \narXiv :2607 .05202v 1 [ cs .AI] 6 Jul 2026  \nAbstract  \nAgent self-evolution in long-horizon LLM systems is largely procedural: useful experience is not merely stored information, but reusable procedures for searching, debugging, and verification. Yet current evaluations do not isolate this form of transfer. Agent benchmarks test single-episode task solving; memory benchmarks target information retention rather than procedural reuse. We introduce EvoAgentBench, a benchmark for agent selfevolution via Ability-guided transfer across four agentic domains: web research, algorithmic reasoning, software engineering, and knowledge work. EvoAgentBench extracts trace-grounded Abilities from agent executions, canonicalizes them into operational units, and builds domain-specific Ability Graphs linking tasks that share procedural overlap. By design, every test task is backed by verified training-side Ability support. Across a 528/267 train/test split, two scaffolds, and three backbones, curated Ability content transfers reliably across model families, but no current automatic method sustains positive gain in all settings. EvoAgentBench shifts self-evolution evaluation from aggregate accuracy comparison to fine-grained diagnosis of experience encoding, routing, and uptake. The benchmark is publicly available at [https://huggingface.co/](https://huggingface.co/)[ ](https://huggingface.co/)datasets/EverMind-AI/EvoAgentBench.  \n1 Introduction  \nA useful agent should not solve every task from scratch. After prior executions, it should retain reusable procedures for searching, debugging, and verification. As LLM agents increasingly master isolated long-horizon tasks (Patwardhan et al., 2025 ; Chen et al., 2025 ; Jain et al., 2024 ; Jimenez et al., 2024), the evaluation frontier is shifting toward exactly this capability: improvement from experience.  \n* Co-corresponding authors.†Project leader.  \nWe study skill-based agent self-evolution, where past trajectories are distilled into external artifacts such as skills, cases, workflows, or evolved prompts, and made available when solving future tasks. This paradigm offers a lightweight, modelagnostic path to improving agent reliability (Shinnet al., 2023 ; Zhao et al., 2024 ; Wang et al., 2025, 2024 ; Fernando et al., 2024) .  \nCurrent evaluations fail to isolate trajectory-toprocedure transfer. Task-centric benchmarks measure whether an agent can solve a held-out task, but they never test whether prior experience converts into reusable procedural knowledge (Patwardhan et al., 2025 ; Chen et al., 2025 ; Jain et al., 2024 ; Jimenez et al., 2024 ; Mialon et al., 2024) . Longcontext and memory benchmarks evaluate recall, retrieval, and personalization over stored information, but stop at information retention, never reaching procedural reuse (Bai et al., 2024 ; Maharana et al., 2024 ; Wu et al., 2025) .  \nBenchmarks for self-evolving agents come closer but still fall short. These evaluate how skill libraries grow over sequential tasks, yet they entangle artifact quality with task order, retrieval policy, and scaffold dynamics (Cai et al., 2025 ; Wu et al., 2024 ; Jiang et al., 2026) . Skill-oriented benchmarks probe procedure quality, but evaluation operates within individual tasks rather than measuring cross-task transfer (L","cbCaibshCJBmozVK","https://ap.wps.com/l/cbCaibshCJBmozVK","pdf",1845031,4,1,15,"English","en",105,"# Abstract\n# Introduction","[{\"question\":\"EvoAgentBench衡量的“能力迁移”具体指什么？\",\"answer\":\"EvoAgentBench将智能体执行轨迹中可验证的Abilities抽取出来，并将其规范化为可操作单元，构建能力图来衡量跨任务的程序级复用，而不是只看是否解题成功或仅保留信息。\"},{\"question\":\"EvoAgentBench如何避免出现“记忆答案或近似重复曝光”导致的虚假提升？\",\"answer\":\"基准遵循“ability-supported yet instance-disjoint”的原则：训练侧提供可迁移的能力支持，但测试任务与训练实例不重复，从而让改进更可能来自程序迁移而非记忆。\"},{\"question\":\"实验结果表明哪些方法或条件可以提升性能？\",\"answer\":\"以能力为依据的参考条件在四个领域均能提升held-out表现，但当前自动自我进化方法仍然较脆弱，增益在不同设置间不一致，并且还观察到负迁移现象（注入的Artifacts可能降低任务表现）。\"}]",1784183913,38,{"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},"evoagentbench-benchmarking-agent-self-evolution-via-ability-transfer","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"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":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":20},"https://docshare.wps.com/document/evoagentbench-benchmarking-agent-self-evolution-via-ability-transfer/82914/",{"url":52,"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","2026-07-23","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},"EvoAgentBench衡量的“能力迁移”具体指什么？","Question",{"text":75,"@type":76},"EvoAgentBench将智能体执行轨迹中可验证的Abilities抽取出来，并将其规范化为可操作单元，构建能力图来衡量跨任务的程序级复用，而不是只看是否解题成功或仅保留信息。","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"EvoAgentBench如何避免出现“记忆答案或近似重复曝光”导致的虚假提升？",{"text":80,"@type":76},"基准遵循“ability-supported yet instance-disjoint”的原则：训练侧提供可迁移的能力支持，但测试任务与训练实例不重复，从而让改进更可能来自程序迁移而非记忆。",{"name":82,"@type":73,"acceptedAnswer":83},"实验结果表明哪些方法或条件可以提升性能？",{"text":84,"@type":76},"以能力为依据的参考条件在四个领域均能提升held-out表现，但当前自动自我进化方法仍然较脆弱，增益在不同设置间不一致，并且还观察到负迁移现象（注入的Artifacts可能降低任务表现）。","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"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":20,"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":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]