[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85686-en":3,"doc-seo-85686-105":29,"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":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},85686,137441390410,"Hazel","https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984",8,"Research & Report","EvoClawBench：Can Agents Learn Reusable Skills from Their Own Runs？","Existing agent benchmarks emphasize task completion, tool use, or the usefulness of predefined/curated skills, but they do not test whether an agent runtime can turn evidence from its own executions into reusable skills that improve later runs. EvoClawBench is proposed as a closed-loop benchmark for repeated, fixture-backed tasks. It compares BASELINE, PRESKILL, and POSTSKILL, covering 100 tasks and 502 sub-problems across coding, data, office, security, operations, and domain documents. Results show learning is selective and cost-sensitive, with strong runtime dependence.","EvoClawBench: Can Agents Learn Reusable Skills from Their Own Runs?  \nZhiyuan Peng1 Xin Yin2 Chenhao Ying1 Zhe Cui3  \nZixiang Ding3 Zhenhua Liu3 Jiang Wu3 Yuan Luo1  \n1 Shanghai Jiao Tong University 2Zhejiang University 3Hithink Research  \n{pzy2000, yingchenhao, [yuanluo}@sjtu.edu.cn](yuanluo}@sjtu.edu.cn)[ ](yuanluo}@sjtu.edu.cn)[xyin@zju.edu.cn](xyin@zju.edu.cn)  \n{cuizhe, dingzixiang, liuzhenhua, [wujiang2}@myhexin.com](wujiang2}@myhexin.com)  \narXiv :2607 .09711v1 [ cs .LG] 23 Jun 2026  \nAbstract  \nExisting agent benchmarks primarily test task completion, tool use, or skill utility, but do not isolate whether a runtime can convert evidence from its own runs into reusable skills that improve fresh executions after authoring overhead. We introduce EvoClawBench, a benchmark for this closed-loop skill-learning question on repeated, fixture-backed tasks. EvoClawBench compares direct execution without skills, PRESKILL authoring before execution, and POSTSKILL summarization from first-run evidence followed by a fresh second execution. The suite contains 100 tasks and 502 sub-problems across coding, data, office, security, operations, and domain-document workflows, with support for multiple agent runtimes. Experiments with OpenClaw and nanobot under local execution show that direct BASELINE performance is strongly runtime-dependent:  \nOpenClaw remains below 20% across models, while nanobot ranges from 56.45% to 96. 13% . Self-authored skills have mixed effects. nanobot GPT-5 .4 stays above 96% in all modes and MiniMax-M2 .7 improves from 90.97% to 94 .50% under POSTSKILL, but nanobot DeepSeek-V4-Pro drops from 77.77% to 4 . 80% with PRESKILL and 0 .99% with POSTSKILL. OpenClaw shows similarly nonmonotonic behavior, with some skill runs near baseline and others collapsing. These results indicate that learning reusable skills from an agent’s own runs is selective and cost-sensitive, rather than an automatic benefit of adding skill authoring to an agent loop.  \n1 Introduction  \nLarge language model agents are increasingly used to turn foundation models into interactive workers. Recent evaluations place agents in repositories, terminals, browsers, graphical interfaces, tool APIs, and executable environments (Fourneyet al., 2024 ; Wang et al., 2025 ; Xu et al., 2025 ;  \nAgashe et al., 2025 ; Research et al., 2026) . Asa result, the benchmark ecosystem has expanded rapidly. SWE-Bench and SWE-agent-style evaluations emphasize repository-level software engineering (Jimenez et al., 2024 ; Yang et al., 2024); Terminal-Bench, AppWorld, WebArena, and OSWorld measure command-line, app-world, browser, and desktop interaction (Zhou et al., 2024 ; Trivediet al., 2024 ; Xie et al., 2024 ; Merrill et al., 2026); and SkillsBench evaluates skills as first-class artifacts across diverse tasks (Li et al., 2026) . These benchmarks have made agent capability more measurable, but they leave a closed-loop learning question under-specified. They primarily test task completion, tool use, environment interaction, or the utility of curated and self-generated skills. They do not isolate the process in which the same agent runtime creates a skill, summarizes evidence from its own completed run, reuses that skill in a fresh execution, and pays the full token, cost, and time overhead of doing so. This leaves a focused evaluation question: Can LLM agents learn reusable skills from their own runs?  \nTo answer this question, we introduce EvoClawBench, a benchmark designed around this question. EvoClawBench evaluates an agent on repeated, structured tasks that contain multiple related subproblems. This design creates an opportunity for the agent to recognize shared patterns and, in the own-run condition, convert first-run evidence into reusable skills. The benchmark focuses on reusable skills because they are becoming a practical way to specialize LLM agents for recurring tool-use tasks. SkillsBench reports 84,192 skills collected within 136 days (Li et al., 2026), su","cbCaitLnYiPFnpS6","https://ap.wps.com/l/cbCaitLnYiPFnpS6","pdf",6149109,1,28,"English","en",105,"# Abstract\n# Introduction\n## Benchmark motivation and gap\n## EvoClawBench design\n## Strategies: BASELINE, PRESKILL, POSTSKILL","[{\"question\":\"EvoClawBench解决了现有智能体基准评测中的什么关键缺口？\",\"answer\":\"现有基准主要评测任务完成、工具使用或既有技能的效用，却没有隔离“同一运行时把自己执行的证据转化为可复用技能，并在新的执行中带来改进”的闭环过程。EvoClawBench专门围绕这个问题设计重复、结构化任务。\"},{\"question\":\"EvoClawBench对每个任务使用了哪些评测策略？\",\"answer\":\"文中比较三种策略：BASELINE（不允许创建/编辑技能直接完成任务）、PRESKILL（先生成任务相关技能再用其完成）、POSTSKILL（先在无技能条件完成，汇总第一轮证据后生成可复用技能，再进行第二轮执行）。\"},{\"question\":\"实验结果表明“从自身运行学习可复用技能”一定会带来收益吗？\",\"answer\":\"不一定。结果显示收益是选择性的且与成本相关，并且强依赖具体运行时/模型：部分模型在某些模式下接近或提升基线，但也出现明显下降甚至崩塌的非单调现象。\"}]",1784205601,71,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"evoclawbench-can-agents-learn-reusable-skills-from-their-own-runs","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/evoclawbench-can-agents-learn-reusable-skills-from-their-own-runs/85686/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","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},"EvoClawBench解决了现有智能体基准评测中的什么关键缺口？","Question",{"text":75,"@type":76},"现有基准主要评测任务完成、工具使用或既有技能的效用，却没有隔离“同一运行时把自己执行的证据转化为可复用技能，并在新的执行中带来改进”的闭环过程。EvoClawBench专门围绕这个问题设计重复、结构化任务。","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"EvoClawBench对每个任务使用了哪些评测策略？",{"text":80,"@type":76},"文中比较三种策略：BASELINE（不允许创建/编辑技能直接完成任务）、PRESKILL（先生成任务相关技能再用其完成）、POSTSKILL（先在无技能条件完成，汇总第一轮证据后生成可复用技能，再进行第二轮执行）。",{"name":82,"@type":73,"acceptedAnswer":83},"实验结果表明“从自身运行学习可复用技能”一定会带来收益吗？",{"text":84,"@type":76},"不一定。结果显示收益是选择性的且与成本相关，并且强依赖具体运行时/模型：部分模型在某些模式下接近或提升基线，但也出现明显下降甚至崩塌的非单调现象。","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":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]