[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-327332-105":59,"doc-detail-327332-en":123},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":116,"head_meta":118,"extra_data":120,"updated_unix":122},105,"en","comfyclaw-self-evolving-skill-harnesses-for-image-generation-workflows","COMFYCLAW - Self-Evolving Skill Harnesses for Image Generation Workflows","","Agents increasingly build workflow systems to help humans complete recurring tasks more efficiently. When workflows repeat within domain-specific settings, agent memory and reusable skills become crucial for recalling patterns, execution constraints, and user preferences from past runs. This work studies workflow-based image generation and introduces COMFYCLAW, an agentic skill evolution harness for controlling ComfyUI workflows, using typed graph editing, staged tools, invalid-edit rollback, and a region-level vision-language verifier to propose actionable repairs.",{"@graph":69,"@context":115},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/comfyclaw-self-evolving-skill-harnesses-for-image-generation-workflows/327332/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/comfyclaw-self-evolving-skill-harnesses-for-image-generation-workflows/327332.png","ImageObject",300,407,{"name":92,"@type":93},"Emma Wilson","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-10","2026-09-21",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",17,{"@type":107,"mainEntity":108},"FAQPage",[109],{"name":110,"@type":111,"acceptedAnswer":112},"What role does the region-level vision-language model (VLM) verifier play?","Question",{"text":113,"@type":114},"It verifies visual outputs, translates failures into actionable repair suggestions, and supports the evolution of reusable skills from verifier feedback across runs.","Answer","https://schema.org",{"og:url":83,"og:type":117,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":119,"canonical":83},"index,follow",{"doc_id":121,"site_id":62},327332,1790302545,{"code":4,"msg":5,"data":124},{"doc_id":121,"user_id":125,"nickname":92,"user_avatar":126,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":127,"file_id":128,"file_url":129,"file_type":130,"file_size":131,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":132,"language":133,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":134,"faqs":135,"seo_title":136,"seo_description":67,"update_tm":137,"read_time":138,"preview_page_count":4},3848291630094,"https://eur-avatar.wpscdn.com/davatar_085a072bc5b1113ac321206ff7593b45","arXiv :2607 .0 1709v 1 [ cs .AI] 2 Jul 2026  \nCOMFYCLAW: Self-Evolving Skill Harnesses for  \nImage Generation Workflows  \nZongxia Li1 ∗ Dawei Liu2 ∗ Fuxiao Liu3 Yuhang Zhou1 Xiyang Wu1 Jingxi Chen 1 Jing Xie 1 Xiaomin Wu 1 Lichao Sun4  \n1University of Maryland 2University of Pennsylvania 3Nvidia 4Lehigh University [zli12321@umd.edu](zli12321@umd.edu) [liudawei@seas.upenn.edu](liudawei@seas.upenn.edu) [lis221@lehigh.edu](lis221@lehigh.edu)  \n§ Code: [https://github.com/Moms-Organic-Agent-Lab/comfyclaw](https://github.com/Moms-Organic-Agent-Lab/comfyclaw)  \nAbstract  \nAgents are increasingly used to construct workflows and help humans complete recurring tasks more efficiently. As these workflows become repeated and domainspecific, agent memory and reusable skills become increasingly important: agents should be able to recall workflow patterns, execution constraints, and user preferences from past runs. We study this problem in workflow-based image generation and introduce COMFYCLAW, an agentic skill evolution harness for controlling ComfyUI workflows. COMFYCLAW represents workflow construction as typed graph editing, exposes tools organized by construction stage, reverts back invalid edits, and uses a region-level vision-language model (VLM) verifier to translate visual failures into actionable repair suggestions. The framework further evolvesa progressively disclosed skill library, where trajectories, execution errors, and verifier feedback from previous runs are distilled into reusable Agent Skills.  \nAcross four benchmark splits, three agent models, and two image backbones, COMFYCLAW achieves the best average image-generation evaluation score across all six agent configurations, outperforming a verifier-only baseline without skill evolution. Human annotations further show that annotators prefer COMFYCLAW over variants without skill evolution. Our results suggest that skill evolution is an effective mechanism for improving agent reliability and performance in recurring visual workflow construction.  \n1 Introduction  \nAgents are moving beyond prompt-only chat interfaces toward workflow execution [1, 2] . This shift is especially important for image generation, where prompt-only interfaces obscure many of the decisions that shape the final result, such as how conditioning is applied, how visual components are composed, and how generated image failures are detected and repaired. Workflow systems such as ComfyUI [3] expose these decisions as editable pipelines, making image generation more inspectable, controllable, and reusable [4, 5] . At the same time, this fine-grained control turns image creation into a workflow problem: an agent must select compatible components, satisfy model constraints, diagnose visual failures, and repair the workflow without breaking execution. Because similar workflow patterns recur across image-generation tasks, effective agents must do more than plan and reflect within a single run: they should also acquire reusable skills from prior executions to avoid repeating the same errors [6, 7] .  \nThis shift changes the agent’s role from specifying prompts to operating an executable procedure. Each step is mediated by a harness that exposes graph edits, runtime feedback, and recovery mechanisms;  \n∗Equal contribution.  \nPreprint.  \none invalid operation can break execution, and some failures are only revealed after the workflow is run [8–13, 4, 14, 15] . Recent workflow-generation agents reduce this burden by generating or refining workflows from natural language [2, 16, 17], while frameworks such as GEMS show promises of closed-loop refinement, memory, and skills for multimodal generation [18] . However, their control interfaces and skill libraries are static, or updated only by passively storing experience as memory rather than actively refining it into reusable skills. Thus, a refinement that succeeds in one run rarely becomes a validated procedure that can be invoked safely in future workflows [8] .  \nTh","cbCaimgHpwpMI71w","https://ap.wps.com/l/cbCaimgHpwpMI71w","pdf",18271161,23,"English","# Introduction\n## Workflow execution vs prompt-only control\n## Skill evolution for reusable procedural knowledge\n## COMFYCLAW framework overview and evaluation","[{\"question\":\"What role does the region-level vision-language model (VLM) verifier play?\",\"answer\":\"It verifies visual outputs, translates failures into actionable repair suggestions, and supports the evolution of reusable skills from verifier feedback across runs.\"}]","COMFYCLAW - Self-Evolving Skill Harnesses for Image Generation Workflows | PDF",1789976998,58]