[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83486-en":3,"doc-seo-83486-105":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},83486,687197100911,"Himbo","https://ap-avatar.wpscdn.com/avatar/a000239b6f1da00475?x-image-process=image/resize,m_fixed,w_180,h_180&k=1782698725881665579",8,"Research & Report","Minos: A Multi-Agent Collaborative Framework for Provenance-Based Backward Tracking","Sophisticated cyber attacks, especially Advanced Persistent Threats (APTs), demand rigorous post-intrusion forensics and the ability to reconstruct attack scenarios. Provenance-based backward tracking traces causality from initial alerts, yet prior techniques often depend on low-level statistics and rigid traversal, missing adversarial intent and suffering from dependency explosion. Minos reframes backward tracking as LLM-driven reasoning with a two-tier multi-agent architecture, improving event assessment, scalable graph exploration, and auditability across diverse datasets.","arXiv :2607 .00440v 1 [ cs .CR] 1 Jul 2026  \nMinos: A Multi-Agent Collaborative Framework for Provenance-Based Backward Tracking  \nJiahui Wang 1 ,2 ,∗ , Zhenyuan Li 1 ,2 ,∗ B , Zhengkai Wang 1 , Xiangmin Shen3 , and  \nFan Zhang 1  \n1 Zhejiang University, Hangzhou, China  \n{wjh_ 13, [lizhenyuan}@zju.edu.cn](lizhenyuan}@zju.edu.cn)  \n2 Ningbo Key Laboratory of Quantum Software and Security, Ningbo, China  \n3 Hofstra University, Hempstead, NY, USA  \nAbstract. Sophisticated cyber attacks, particularly Advanced Persistent Threats (APTs), necessitate rigorous post-intrusion forensic analysis.  \nProvenance-based backward tracking serves as a pivotal capability for reconstructing attack scenarios by tracing causality from initial alerts.  \nHowever, existing methods frequently rely on low-level statistical features and rigid traversal strategies. These approaches fail to capture highlevel adversarial intent, especially against stealthy living-off-the-land techniques, and inevitably struggle with “dependency explosion”.  \nTo address these challenges, we propose Minos, a multi-agent collaborative framework that reconceptualizes backward tracking as a Large Language Model (LLM)-driven reasoning process. Minos operates via a two-tiered architecture. For individual event assessment, it introduces a structured framework to overcome the inherent limitations of LLMs: it employs a hierarchical context model for persistent state maintenance, implements retrieval-augmented reasoning with citation verification to ground inferences, and incorporates an adversarial deliberation mechanism to mitigate sycophancy bias. For end-to-end graph exploration, Minos orchestrates four specialized agents under a finite state machine (FSM), replacing exhaustive topological traversal with hypothesis-guided reasoning and “count-first” query protocols to prune the search space.  \nComprehensive evaluations on 14 attack scenarios across five public datasets demonstrate that Minos achieves average recall and precision of 0.92 and 0.64, respectively, significantly outperforming state-of-the-art baselines while generating attack subgraphs that are 49% more compact.  \nFurthermore, Minos generates interpretable reasoning at every step, providing robust support for auditing and system refinement. Ultimately, our exploration validates the efficacy of leveraging LLMs for automated provenance-based backward tracking.  \nKeywords: Backward Tracking · Provenance Analysis · Large Language Model · Multi-Agent System  \n∗ Two authors contribute equally to this work.  \nB Corresponding author: [lizhenyuan@zju.edu.cn](lizhenyuan@zju.edu.cn)  \n2 J. Wang et al.  \n1 Introduction  \nSophisticated cyber attacks, particularly Advanced Persistent Threats (APTs), pose an escalating threat to critical infrastructures. As adversaries frequently evade initial defenses to establish prolonged persistence within compromised networks, the ability to reconstruct a comprehensive attack scenario after anomaly detection becomes a critical forensic capability. To support such post-intrusion analysis, provenance graphs [18] have emerged as instrumental tools: by parsing kernel-level audit logs into a unified graph representation where nodes represent system entities and edges capture causal interactions, they transform discrete, fragmented log entries into a temporal graph with inherent causality, enabling systematic investigation. The central objective in this investigation is backward tracking: starting from a Point-of-Interest (POI) event flagged by an Intrusion Detection System (IDS), analysts trace backward along the provenance graph toreconstruct the adversarial operations and locate the attack entry points [16] .  \nPrior research has advanced backward tracking through several strategies, including reachability analysis [14 ,9], statistical anomaly detection [13], and semantic clustering [32] . Despite their contributions, existing approaches still encounter two critical challenges. First, they often r","cbCaihbOwGG7ldsa","https://ap.wps.com/l/cbCaihbOwGG7ldsa","pdf",2007466,4,1,20,"English","en",105,"# Introduction\n## Provenance graphs and backward tracking\n## Challenges in existing methods\n## LLM-based motivations\n## Proposed Minos framework","[{\"question\":\"What problem does Minos address in provenance-based backward tracking?\",\"answer\":\"Minos targets two main issues: existing methods often miss high-level adversarial intent and they suffer from dependency explosion during graph traversal. It reframes backward tracking as LLM-driven reasoning to improve both effectiveness and scalability.\"},{\"question\":\"How does Minos improve individual event assessment?\",\"answer\":\"For each event, Minos uses a structured framework to overcome generic LLM limitations, including a hierarchical context model for persistent state, retrieval-augmented reasoning with citation verification, and an adversarial deliberation mechanism to reduce sycophancy bias.\"},{\"question\":\"How does Minos perform end-to-end graph exploration efficiently?\",\"answer\":\"Minos orchestrates four specialized agents using a finite state machine, replacing exhaustive topological traversal with hypothesis-guided reasoning and count-first query protocols that prune the search space.\"}]",1784188349,50,{"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},"minos-a-multi-agent-collaborative-framework-for-provenance-based-backward-tracking","",{"@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/minos-a-multi-agent-collaborative-framework-for-provenance-based-backward-tracking/83486/",{"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","2026-07-24","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 Minos address in provenance-based backward tracking?","Question",{"text":75,"@type":76},"Minos targets two main issues: existing methods often miss high-level adversarial intent and they suffer from dependency explosion during graph traversal. It reframes backward tracking as LLM-driven reasoning to improve both effectiveness and scalability.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does Minos improve individual event assessment?",{"text":80,"@type":76},"For each event, Minos uses a structured framework to overcome generic LLM limitations, including a hierarchical context model for persistent state, retrieval-augmented reasoning with citation verification, and an adversarial deliberation mechanism to reduce sycophancy bias.",{"name":82,"@type":73,"acceptedAnswer":83},"How does Minos perform end-to-end graph exploration efficiently?",{"text":84,"@type":76},"Minos orchestrates four specialized agents using a finite state machine, replacing exhaustive topological traversal with hypothesis-guided reasoning and count-first query protocols that prune the search space.","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},[93,97,101,105,110,114,119,122,126,129,133],{"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":29,"slug":113},6,"Technology","technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"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":22,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":22,"doc_module":4,"doc_module_name":46,"category_name":127,"show_sort_weight":22,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":46,"category_name":135,"show_sort_weight":106,"slug":136},19,"General","general"]