[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-81734-en":3,"doc-seo-81734-105":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},81734,4810365810221,"Aurora","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","ATM: CID-Brokered Pre-Write Admission for Multi-Agent Code Co-Synthesis","Multi-agent LLM systems can split software engineering into planning, generation, validation, and repair, but still face a gap before any governed shared mutation happens: deciding which write intents run in parallel, which need deterministic composition or serialization, and which must fail closed. The AI-Atomic-Framework (ATM) provides specification-grounded governance in a single domain. ATM links intent, scope, admission, validation, and evidence obligations through a CID broker, using adapter-guided atomization, parallel admission routing, and a neutral steward for applying writes.","arXiv :2607 .00041v1 [ cs . SE] 29 Jun 2026  \nATM: CID-Brokered Pre-Write Admission for Multi-Agent Code Co-Synthesis  \nA Specification-Grounded Governance Substrate for Software Agents  \nEagl Huang  \n[eaglhuang@gmail.com](eaglhuang@gmail.com)  \nAbstract  \nMulti-agent LLM systems can decompose software-engineering tasks into planning, generation, validation, and repair. A narrower system gap remains: after several agents have formed write intents within the same controlled filesystem, worktree, or service domain, but before any governed shared mutation is applied, the system must determine which intents may proceed in parallel, which require deterministic composition or serialization, and which must take a fail-closed path. To address this gap, we present the AI-Atomic-Framework (ATM), a specification-to-evidence governance substrate for software agents operating within a single governance domain.  \nATM binds task intent, repository scope, write admission, validation, and evidence obligations into a single governance chain. Within this chain, the Content Identifier (CID) broker serves as the shared-mutation admission subsystem. Adapter-guided atomization maps write intents to semantic atoms and bounded regions; the broker then routes them to parallel admission, deterministic composition, serialization, or fail-closed refinement. When the persistent atom map is incomplete, virtual atoms provide temporary, auditable governance units that preserve bounded-region comparability. Governed shared writes are ultimately applied by a neutral steward rather than directly by the proposing agents.  \nEvaluation combines controlled, field, adoption, and extension evidence. Controlled evidence includes a 12-scenario deterministic design matrix, three archived runner cases, and ATM-AdmissionBench, in which v0.1 freezes the benchmark substrate and v0.2 provides the paper-facing profile over 20 unique scenarios and 42 mode-level comparisons. The policy, ablation, adversarial, and enforcement rows are derived views of those scenarios rather than independent population samples. Field evidence comprises three archived same-file boundary cases–POS2, B-12, and BLOCK–while a three-week external-adopter study, observations of batch scheduling and CID stability, and an operational recovery-routing benchmark provide complementary evidence of operability, recoverability, and runtime transparency. Taken together, these results support feasibility within the observed single-domain settings, but not broad comparative superiority over alternative concurrency-control systems. ATM neither replaces Git merge nor addresses cross-clone or cross-PR governance; its claim is limited to an implementable, auditable, and incrementally extensible pre-write admission layer for governed shared mutations.  \n1 Introduction  \n1.1 Motivation  \nThe central question of this paper is not whether individual agents can produce usable code, but whether, once multiple agents have formed write intents within the same controlled filesystem, worktree, or service domain, the system can determine, before any governed shared mutation is applied, which intents may proceed in parallel, which require deterministic composition or serialization, and which must be blocked. ATM frames this as a pre-write admission problem within a single governance domain. Same-file edits, shared registries, generated artifacts, and task-state machines are treated as different manifestations of the same governance boundary rather than as separate claim scopes. This paper does not address cross-machine clones, remote branches, or PR-level distributed coordination; it focuses on auditable admission decisions made before governed shared-state mutation. This setting lies within the broader LLM-for-SE landscape. The systematic review by Hou et al. [31] divides LLM applications in software engineering into code generation, testing, maintenance, and coordination, and identifies multi-agent coordination as an emerging an","cbCaikJOemS8aB1Z","https://ap.wps.com/l/cbCaikJOemS8aB1Z","pdf",392124,3,1,40,"English","en",105,"# Introduction\n## Motivation\n## Background and related work","[{\"question\":\"What problem does ATM address in multi-agent code co-synthesis?\",\"answer\":\"ATM focuses on the decision step before any governed shared mutation: determining which agents’ write intents can proceed in parallel, must be deterministically composed or serialized, or must be blocked via fail-closed refinement.\"},{\"question\":\"How does ATM enforce governance for shared mutations?\",\"answer\":\"ATM ties task intent, repository scope, write admission, validation, and evidence into one governance chain. A CID broker acts as the shared-mutation admission subsystem, routing atomized write intents to parallel admission, deterministic composition, serialization, or fail-closed refinement, while a neutral steward applies governed writes.\"},{\"question\":\"What does the evaluation show about ATM’s effectiveness?\",\"answer\":\"Evaluation combines deterministic design-matrix controlled evidence, archived runner and boundary cases, and external adopter evidence on batch scheduling and CID stability. Results support feasibility in observed single-domain settings, while ATM does not claim broad superiority over other concurrency-control approaches.\"}]",1784175712,101,{"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},"atm-cid-brokered-pre-write-admission-for-multi-agent-code-co-synthesis","",{"@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/atm-cid-brokered-pre-write-admission-for-multi-agent-code-co-synthesis/81734/",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","2026-07-23","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},"What problem does ATM address in multi-agent code co-synthesis?","Question",{"text":75,"@type":76},"ATM focuses on the decision step before any governed shared mutation: determining which agents’ write intents can proceed in parallel, must be deterministically composed or serialized, or must be blocked via fail-closed refinement.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does ATM enforce governance for shared mutations?",{"text":80,"@type":76},"ATM ties task intent, repository scope, write admission, validation, and evidence into one governance chain. A CID broker acts as the shared-mutation admission subsystem, routing atomized write intents to parallel admission, deterministic composition, serialization, or fail-closed refinement, while a neutral steward applies governed writes.",{"name":82,"@type":73,"acceptedAnswer":83},"What does the evaluation show about ATM’s effectiveness?",{"text":84,"@type":76},"Evaluation combines deterministic design-matrix controlled evidence, archived runner and boundary cases, and external adopter evidence on batch scheduling and CID stability. Results support feasibility in observed single-domain settings, while ATM does not claim broad superiority over other concurrency-control approaches.","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},[93,97,101,105,110,115,119,122,127,130,134],{"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":22,"slug":118},7,"Healthcare","healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]