Datasets:

fineset-io
/

llm-agent-papers

id string	sources list	title string	abstract string	authors list	categories list	fields_of_study list	published_date timestamp[s]	url string	pdf_url string	arxiv_id string	doi string	citation_count int64	has_code bool	code_url string	venue string	quality_score float64
f1e9a3cd0534fb97e149ba30818ae576fc64bd509e69b7c64fad1fda368ee3de	[ "arxiv" ]	Exploration Structure in LLM Agents for Multi-File Change Localization	Software engineering tools increasingly rely on LLM based agents to localize files to change to resolve a software issue. Most AI agents explore repositories linearly, that is, visiting one directory or file per step. We postulate that this is a structural mismatch for changes that span several subsystems. We compare l...	[ "Akeela Darryl Fattha", "Kia Ying Chua", "Lingxiao Jiang", "Laura Wynter" ]	[ "cs.SE", "cs.AI" ]	[]	2026-06-10T00:00:00	https://arxiv.org/abs/2606.11976	https://arxiv.org/pdf/2606.11976v1	2606.11976	null	0	false	null	null	0
ae195c5a11ce9cd1c2f521b98fc8e6bf9ec4fadb23bbeed5453b2cc19ae1f643	[ "arxiv", "semantic_scholar" ]	Early-Token Confidence Predicts Reasoning Quality in Multi-Agent LLM Debate	Evaluating reasoning quality in multi-agent LLM systems is challenging, especially for open-ended tasks without reference answers. We investigate whether intrinsic confidence signals, token-level log-probabilities from decoding, can predict reasoning quality as assessed by LLM-as-judge evaluation. Using a debate-based ...	[ "Ali Keramati", "Justin Cheok", "Jacob Horne", "Mark Warschauer" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-09T00:00:00	https://arxiv.org/abs/2606.10307	https://arxiv.org/pdf/2606.10307v1	2606.10307	null	0	false	null	null	0
2093031762620b5e6328a41fd02c5910fd3b0fac9aaae4e1bba4fe5c01f52017	[ "arxiv", "semantic_scholar" ]	Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation	Large language model (LLM) agents are rapidly moving from conversational interfaces to software components that plan, invoke tools, maintain memory, and act on external environments. This transition changes the nature of security risk. In agentic settings, failures are no longer limited to unsafe text generation. Untru...	[ "Yuchen Ling", "Shengcheng Yu", "Zhenyu Chen", "Chunrong Fang" ]	[ "cs.CR", "cs.AI" ]	[ "Computer Science" ]	2026-06-09T00:00:00	https://arxiv.org/abs/2606.10749	https://arxiv.org/pdf/2606.10749v1	2606.10749	null	0	false	null	null	0
1175f28a811a7734352b0aa219624996906b3d4783a4d49d48c4cff8f15b7271	[ "arxiv", "semantic_scholar" ]	Game-Theoretic Multi-Agent Control for Robust Contextual Reasoning in LLMs	Large Language Models (LLMs) in multi-turn interactions maintain evolving context rather than generating isolated responses, making them vulnerable to prompt-injection and context-poisoning attacks in which locally plausible adversarial fragments gradually distort reasoning trajectories. Existing defenses mainly filter...	[ "Saeid Jamshidi", "Amin Nikanjam", "Arghavan Moradi Dakhel", "Kawser Wazed Nafi", "Foutse Khomh" ]	[ "cs.CR", "cs.MA" ]	[ "Computer Science" ]	2026-06-09T00:00:00	https://arxiv.org/abs/2606.10322	https://arxiv.org/pdf/2606.10322v1	2606.10322	null	0	false	null	null	0
19304d81686c86a0104aa49b91a6852644e0447736571e9f2f6bc8f0f65422a4	[ "arxiv", "semantic_scholar" ]	The Confident Liar: Diagnosing Multi-Agent Debate with Log-Probabilities and LLM-as-Judge	Multi-agent debate systems are typically evaluated only on whether the final answer is correct, overlooking the quality of the intermediate reasoning that debate is designed to produce. This paper studies the relationship between three signals in multi-agent debate: token-level log-probability distributions over reason...	[ "Ali Keramati", "Justin Cheok", "Jacob Horne", "Mark Warschauer" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-06-09T00:00:00	https://arxiv.org/abs/2606.10296	https://arxiv.org/pdf/2606.10296v1	2606.10296	null	0	false	null	null	0
8e1008b62ae6b0d13f4e09bddf01d2b8ea9e59a46c80e66193e106378e554d22	[ "arxiv", "semantic_scholar" ]	The Arbiter Agent: Continually Monitoring Multi-Agent Conversations to Detect Emergent Misalignment	As AI systems built from multiple language-model agents become more common, they are increasingly used to make decisions together: discussing, negotiating, and acting on shared tasks. While individual agents may appear well-aligned when tested on their own, problems can arise from how they interact with one another. We...	[ "Filippo Tonini", "Federico Torrielli", "Anton Danholt Lautrup", "Peter Schneider-Kamp", "Mustafa Mert Çelikok", "Lukas Galke Poech" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-09T00:00:00	https://arxiv.org/abs/2606.10747	https://arxiv.org/pdf/2606.10747v1	2606.10747	null	0	true	https://github.com/aisilab/arbiter	null	0
b9d5e324ad7e57f2dbcf0e7ea938d961dffbee982adb1e73e75a2d87bb286c54	[ "arxiv" ]	INFRAMIND: Infrastructure-Aware Multi-Agent Orchestration	Existing multi-agent LLM orchestration methods, ranging from brute-force ensembles to learned routers, select models and topologies based on task and model features. However, these methods do not consider the runtime state of the serving infrastructure. On shared GPU clusters under concurrent load, this infrastructure ...	[ "Ahasan Kabir", "Jiaqi Xue", "Mengxin Zheng", "Qian Lou" ]	[ "cs.AI" ]	[]	2026-06-09T00:00:00	https://arxiv.org/abs/2606.11440	https://arxiv.org/pdf/2606.11440v1	2606.11440	null	0	false	null	null	0
70c2f0d8aded166fe1a6201c70f204817780c4547c6900c10d25b731dd1bd260	[ "arxiv", "semantic_scholar" ]	Divide and Cooperate: Role-Decomposed Multi-Agent LLM Training with Cross-Agent Learning Signals	Modern language agents which perform multi-step reasoning have shown strong performance in knowledge-intensive question answering. However, existing approaches typically couple evidence acquisition and answer generation within a single policy. This forces a single model to play multiple potentially conflicting roles, i...	[ "Jaewan Park", "Solbee Cho", "Jay-Yoon Lee" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-06-09T00:00:00	https://arxiv.org/abs/2606.10684	https://arxiv.org/pdf/2606.10684v1	2606.10684	null	0	false	null	null	0
c6a5773175161ea0251d226cb68e6e144ec0a651216ccbcb4cd0a6b832307b5d	[ "arxiv", "semantic_scholar" ]	Context-Fractured Decomposition Attacks on Tool-Using LLM Agents: Exploiting Artifact Provenance Gaps	Tool-using LLM agents interact with the world through actions that persist state in artifacts (e.g., workspace files or logs). Consequently, jailbreak defenses must reason about cross-step composition rather than isolated text. Yet most existing attacks and defenses, including ``multi-turn'' jailbreaks such as Crescend...	[ "Xiaofeng Lin", "Yukai Yang", "Daniel Guo", "Sahil Arun Nale", "Charles Fleming", "Guang Cheng" ]	[ "cs.CR", "cs.AI" ]	[ "Computer Science" ]	2026-06-08T00:00:00	https://arxiv.org/abs/2606.09084	https://arxiv.org/pdf/2606.09084v1	2606.09084	null	0	false	null	null	0
1024a29f8c6411aa5434ac1b840eaf327f14b2912c1789702afcce0562bff4da	[ "arxiv", "semantic_scholar" ]	AGENTSERVESIM: A Hardware-aware Simulator for Multi-Turn LLM Agent Serving	Multi-turn LLM agents interleave model calls with external tool invocations, shifting serving from stateless request processing to stateful program execution. Serving these workloads requires scheduling, KV-cache management, and routing policies that use program-level context, including turn dependencies, tool-induced ...	[ "Rakibul Hasan Rajib", "Mengxin Zheng", "Qian Lou" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-06-08T00:00:00	https://arxiv.org/abs/2606.09613	https://arxiv.org/pdf/2606.09613v1	2606.09613	null	0	false	null	null	0
9f832e79793f1899478661987639471a4c7fe28bbe918d4dfc029dc0af44c369	[ "arxiv", "semantic_scholar" ]	Prisma-World: Camera-Controllable Multi-Agent Video World Model	Video world models have made rapid progress in generating controllable visual experiences, but most of them still simulate the world from a single observer. Extending such models to multiple agents raises a central challenge: if each agent's future state is generated independently, overlapping views may instantiate dif...	[ "Huiqiang Sun", "Zhan Peng", "Size Wu", "Kun Wang", "Kang Liao", "Dianyi Wang", "Xingyu Zeng", "Sheng Jin", "Yangguang Li", "Zhiguo Cao", "Ziwei Liu", "Wei Li" ]	[ "cs.CV" ]	[ "Computer Science" ]	2026-06-08T00:00:00	https://arxiv.org/abs/2606.09507	https://arxiv.org/pdf/2606.09507v1	2606.09507	null	0	false	null	null	0
88f5e95d2ddcad1d198398171bdc862c63afe907e41fe0157736e2e54cb9fa02	[ "arxiv", "semantic_scholar" ]	PerspectiveGap: A Benchmark for Multi-Agent Orchestration Prompting	Real-world LLM applications are moving beyond single-agent workflows toward orchestrated multi-agent systems, yet current models still struggle to determine what each sub-agent needs to know. To measure this, we introduce PerspectiveGap, a benchmark for evaluating LLMs' ability to compose orchestration prompts for mult...	[ "Youran Sun", "Xingyu Ren", "Kejia Zhang", "Xinpeng Liu", "Jiaxuan Guo" ]	[ "cs.CL", "cs.MA" ]	[ "Computer Science" ]	2026-06-07T00:00:00	https://arxiv.org/abs/2606.08878	https://arxiv.org/pdf/2606.08878v1	2606.08878	null	0	false	null	null	0
4422ba6bd5afbbaeee25db98661efd55f326124dd1123c3660245ec40556ab64	[ "arxiv", "semantic_scholar" ]	SGTO-MAS: Secure Gorilla Troops Optimization for Multi-Agent LLM Systems	Multi-agent large language model (LLM) systems offer strong capabilities for complex reasoning and decision-making, yet coordination across agents introduces error propagation, security risks, and inefficient use of resources. Existing methods often rely on heuristic, static strategies and lack a principled mechanism f...	[ "Saeid Jamshidi" ]	[ "cs.CR" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.07940	https://arxiv.org/pdf/2606.07940v1	2606.07940	null	0	false	null	null	0
a67b489174115d59302461692a661b33a73183bda3dc6ee38b4d8c55136bf0d4	[ "arxiv", "semantic_scholar" ]	Causal Agent Replay: Counterfactual Attribution for LLM-Agent Failures	When an LLM agent fails -- issues a refund it should not have, calls the wrong tool, leaks data -- existing tooling answers what happened (observability) or whether it passed (evaluation), but not which step caused the failure. The obvious heuristics are wrong: the step that executes the harmful action is usually not t...	[ "Jaineet Shah" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.08275	https://arxiv.org/pdf/2606.08275v1	2606.08275	null	0	true	https://github.com/jaineet17/causal-agent-replay	null	0
8fd6fafdc2599ecc0966046ca9214783e084f4345d20359bf74155d5fc602f0c	[ "arxiv", "semantic_scholar" ]	Hallucination Cascade: Analyzing Error Propagation in Multi-Agent LLM Systems	Large Language Models (LLMs) generate fluent text but remain vulnerable to hallucinations, producing unsupported, inconsistent, and factually incorrect claims. Most prior work treats hallucination as a static property of isolated outputs. In multi-agent LLM systems, however, responses are exchanged across agents, revis...	[ "Saeid Jamshidi", "Arghavan Moradi Dakhel", "Kawser Wazed Nafi", "Foutse Khomh" ]	[ "cs.CR" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.07937	https://arxiv.org/pdf/2606.07937v1	2606.07937	null	0	false	null	null	0
a1ff74635280f78bb7a28de1c484cf0245d0ee85ca50f73e05833318704c31ef	[ "arxiv", "semantic_scholar" ]	Toward Human-Centered Multi-Agent Systems: Integrating Cognition, Culture, Values, and Cooperation in AI Agents	The emergence of large language model (LLM)-based agents and multi-agent systems has enabled a shift from narrow task automation to more autonomous decision-making. Despite progress in language generation, planning, tool use, and coordination, most agents still treat intelligence as prediction, optimization, and task c...	[ "Safia Baloch", "Rahemeen Khan" ]	[ "cs.MA" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.08274	https://arxiv.org/pdf/2606.08274v1	2606.08274	null	0	false	null	null	0
3e9b0be011efef4291109bf370698c6756c5b0c499c1adfd06004cb7e245dd2e	[ "arxiv", "semantic_scholar" ]	Collective Hallucination in Multi-Agent LLMs:Modeling and Defense	Hallucinations in large language models (LLMs) create heightened risks in multi-agent settings, where recursive agent interactions can propagate, reinforce, and amplify unsupported claims. This paper models hallucination as a system-level, time-evolving process across a network of interacting LLM agents, where nodes re...	[ "Saeid Jamshidi" ]	[ "cs.CR" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.07941	https://arxiv.org/pdf/2606.07941v1	2606.07941	null	0	false	null	null	0
0474cd39b1aa7342093d0e6062f33a01f5c7dfee901afcd1a6609e30052d07e3	[ "arxiv", "semantic_scholar" ]	Benchmarking Open-Ended Multi-Agent Coordination in Language Agents	As language models are increasingly deployed as autonomous agents, they must coordinate with others over long horizons in open-ended interactive tasks. Yet existing evaluations rarely test these demands together, instead emphasising single-agent tasks, short interactions, or highly structured multi-agent settings. We i...	[ "Kale-ab Abebe Tessera", "Andras Szecsenyi", "Cameron Barker", "Alexander Rutherford", "Davide Paglieri", "Aidan Scannell", "Henry Gouk", "Elliot J. Crowley", "Tim Rocktäschel", "Amos Storkey" ]	[ "cs.AI", "cs.LG", "cs.MA" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.08340	https://arxiv.org/pdf/2606.08340v1	2606.08340	null	0	true	https://github.com/alem-world/alem-env	null	0
0f34a99d9b4d1f3761b22d5da920010273bb0d32ea3375cb6cd6f2ec6095fd5b	[ "arxiv", "semantic_scholar" ]	Emergence World: A Platform for Evaluating Long-Horizon Multi-Agent Autonomy	Most evaluations of LLM agents look like exams: a discrete task, a clean environment, a score in minutes or hours. We argue that this approach is mismatched with the deployment conditions of autonomous systems, where the relevant timescale can be weeks to months, and where the dynamics that matter most, such as behavio...	[ "Deepak Akkil", "Ravi Kokku", "Karthik Vikram", "Tamer Abuelsaad", "Aditya Vempaty", "Satya Nitta" ]	[ "cs.MA", "cs.AI" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.08367	https://arxiv.org/pdf/2606.08367v1	2606.08367	null	0	false	null	null	0
c285e2cada1775bd86d02c7c4def51cdbb8a1e608e41c3d06054141b215edc5b	[ "arxiv", "semantic_scholar" ]	"So There's a Catch-22 Here": How Early Adopters Who Build Multi-Agent LLM Systems Conceptualize Transparency	Multi-agent large language model (LLM) systems are rapidly emerging, yet transparency, a cornerstone of responsible AI, remains under-defined in these distributed architectures, which have complexities of inter-agent coordination and orchestration. In this paper, we present one of the first empirical study of how early...	[ "Suchismita Naik", "Samir Passi", "Mihaela Vorvoreanu", "Scott Saponas", "Amanda Hall" ]	[ "cs.HC", "cs.AI" ]	[ "Computer Science" ]	2026-06-06T00:00:00	https://arxiv.org/abs/2606.08323	https://arxiv.org/pdf/2606.08323v1	2606.08323	null	0	false	null	null	0
78ffc6d5ce47ae083029340c81d5498b49d5f086d452d546711feb85dfe4dc77	[ "arxiv", "semantic_scholar" ]	Contract2Tool: Learning Preconditions and Effects for Reliable Tool-Augmented LLM Agents	Tool-augmented large language model agents increasingly rely on external APIs, but standard tool schemas describe how to call a tool, not when the tool is causally appropriate or what task state it produces. Causal tool filtering addresses this gap by using lightweight contracts that specify each tool's preconditions, ...	[ "Rahul Suresh Babu", "Laxmipriya Ganesh Iyer" ]	[ "cs.AI", "cs.SE" ]	[ "Computer Science" ]	2026-06-05T00:00:00	https://arxiv.org/abs/2606.07904	https://arxiv.org/pdf/2606.07904v1	2606.07904	null	0	false	null	null	0
024194d091c59bda910a8e4920c8bbfef34f2cd1739e70826f21985b0b57c2d9	[ "arxiv", "semantic_scholar" ]	Exploring Agentic Tool-Calling Decisions via Uncertainty-Aligned Reinforcement Learning	Large language model (LLM)-based agents often make suboptimal tool-use decisions, including unsupported tool invocation and hallucinated direct responses, which may accumulate errors throughout multi-step interactions. Existing approaches mainly improve these behaviors through inference-time correction or coarse-graine...	[ "Yijin Zhou", "Linqian Zeng", "Xiaoya Lu", "Wenyuan Xie", "Dongrui Liu", "Junchi Yan", "Jing Shao" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-05T00:00:00	https://arxiv.org/abs/2606.06976	https://arxiv.org/pdf/2606.06976v1	2606.06976	null	0	false	null	null	0
043e1592190460ae552cab9ee16ee4ecc9b5e2d347d349ecb8f55290f1cc5413	[ "arxiv", "semantic_scholar" ]	Does Persona Make LLMs K-pop Fans? A Pilot Study of LLM-Based Online Concert Audience Agents	A concert is a collective experience, but recorded performance videos are typically watched alone, stripping away the shared audience presence that makes concerts feel eventful. We investigate whether persona-based LLM audience agents can recreate aspects of this collective experience by generating real-time fan chat a...	[ "Kirak Kim", "Hyojin Kim", "Yejin Son", "Sungyoung Kim", "Kyung Myun Lee" ]	[ "cs.HC", "cs.AI" ]	[ "Computer Science" ]	2026-06-05T00:00:00	https://arxiv.org/abs/2606.07837	https://arxiv.org/pdf/2606.07837v1	2606.07837	null	0	false	null	null	0
d663a17dc06a3db96ec0f80fe501f6ab78deb97b0add589d55c70f0b4f4914d7	[ "arxiv", "semantic_scholar" ]	PDE-Agents: An LLM-Orchestrated Multi-Agent Framework for Automated Finite Element Simulations with Knowledge Graph-Augmented Reasoning	We present PDE-Agents, a multi-agent ecosystem that automates the full lifecycle of partial differential equation (PDE) / finite element method (FEM) simulations through natural-language interaction. Three specialist large language model (LLM) agents (Simulation, Analytics, Database) are orchestrated via a LangGraph su...	[ "Sayan Adhikari", "Gulshan Noorsumar", "Øyvind Jensen" ]	[ "physics.comp-ph", "math-ph" ]	[ "Physics", "Mathematics" ]	2026-06-05T00:00:00	https://arxiv.org/abs/2606.07850	https://arxiv.org/pdf/2606.07850v1	2606.07850	null	0	true	https://github.com/MatPro-IFE/pde-agents	null	0
5ea3182923a447700ff7de9a75dc9d912a92dbc16a5d4fdefccb458d42d9e167	[ "arxiv", "semantic_scholar" ]	YouZhi: Towards High-Concurrency Financial LLMs via Adaptive GQA-to-MLA Transition	Large language models (LLMs) drive significant financial innovations, yet their high-concurrency deployment is severely bottlenecked by KV cache memory overhead, which inflates infrastructure costs and throttles scalability. To address this, we propose YouZhi-LLM, a highly efficient financial LLM empowered by a compreh...	[ " PSBC LLM Team", " Huawei LLM Team", "Ruihan Long", "Junjie Wu", "Tianan Zhang", "Duo Zhang", "Yaozong Wu", "Jinbin Fu", "Chang Liu", "Zhentao Tang", "Wenshuang Yang", "Xin Wang", "Zhihao Song", "Ning Huang", "Wenjing Xu", "Shuai Zong", "Shupei Sun", "Sen Wang", "Jing Hu", "Bi...	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.05868	https://arxiv.org/pdf/2606.05868v1	2606.05868	null	0	false	null	null	0
5df83aefc7ef75e9f29c217c072de2a4bb944d9f7ffa61b3c00591b44f8a6dd8	[ "arxiv", "semantic_scholar" ]	CollabSim: A CSCW-Grounded Methodology for Investigating Collaborative Competence of LLM Agents through Controlled Multi-Agent Experiments	Multi-agent systems (MAS) built on large language models have shown growing promise, with their effectiveness resting on agents' ability to coordinate through text-based channels much as human teams do. Yet recent study suggests that MAS often falter not because agents lack individual task-solving ability, but because ...	[ "Jiaju Chen", "Bo Sun", "Yuxuan Lu", "Yun Wang", "Dakuo Wang", "Bingsheng Yao" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.06399	https://arxiv.org/pdf/2606.06399v2	2606.06399	null	0	false	null	null	0
b01f83b507b02718e0789e4098d33bac3a7760e3ba45687731ec86d23635d19a	[ "arxiv", "semantic_scholar" ]	ToolChoiceConfusion: Causal Minimal Tool Filtering for Reliable LLM Agents	Large language model agents increasingly rely on external tools, but larger tool menus can reduce reliability and efficiency by increasing wrong-tool calls, premature actions, and token cost. Existing tool-selection methods often optimize semantic relevance, exposing tools whose names or descriptions match the user req...	[ "Rahul Suresh Babu", "Laxmipriya Ganesh Iyer" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.06284	https://arxiv.org/pdf/2606.06284v1	2606.06284	null	1	false	null	null	0.0753
854499a642e3f076e703ef4c20a5c9e9b76a6e1c7ff57e2128bf9047e8ef2556	[ "arxiv", "semantic_scholar" ]	Do More Agents Help? Controlled and Protocol-Aligned Evaluation of LLM Agent Workflows	Does adding more agents help an LLM workflow once compared systems share the same benchmark loader, tool access, answer contract, usage accounting, and trajectory logging? We introduce BenchAgent, an evaluation framework that places single-agent, fixed multi-agent (MAS), and evolving MAS workflows under one normalized ...	[ "Yuhang Fu", "Ruishan Fang", "Jiaqi Shao", "Huiyu Zheng", "Zhengtao Zhu", "Bing Luo", "Tao Lin" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.05670	https://arxiv.org/pdf/2606.05670v1	2606.05670	null	0	true	https://github.com/LINs-lab/MASArena/tree/BenchAgent	null	0
7c7eabd773195d897701dd3192badd6d69ee4bb2a6265661ea5803bb2192dfea	[ "arxiv", "semantic_scholar" ]	Autoregressive Diffusion World Models for Off-Policy Evaluation of LLM Agents	Evaluating large language model (LLM) agents in multi-turn interactive environments is expensive and risky, as it requires online environment interaction. We propose ADWM (Autoregressive Diffusion World Model), an evaluation framework that estimates the performance of a new LLM agent policy purely from pre-collected tr...	[ "Kaixuan Liu", "Guojun Xiong", "Weinan Zhang", "Shengpu Tang" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.05558	https://arxiv.org/pdf/2606.05558v1	2606.05558	null	0	false	null	null	0
35d10bc510ab3a0e127baf9206fd9469245c3c7008336224b1097ce176debeec	[ "arxiv", "semantic_scholar" ]	The Meta-Agent Challenge: Are Current Agents Capable of Autonomous Agent Development?	Current AI benchmarks evaluate agents on task execution within human-designed workflows. These evaluations fundamentally fail to measure a critical next-level capability: whether models can autonomously develop agent systems. We introduce the Meta-Agent Challenge (MAC), an evaluation framework designed to test the capa...	[ "Xinyu Lu", "Tianshu Wang", "Pengbo Wang", "zujie wen", "Zhiqiang Zhang", "Jun Zhou", "Boxi Cao", "Yaojie Lu", "Hongyu Lin", "Xianpei Han", "Le Sun" ]	[ "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-06-03T00:00:00	https://arxiv.org/abs/2606.04455	https://arxiv.org/pdf/2606.04455v1	2606.04455	null	0	true	https://github.com/ant-research/meta-agent-challenge	null	0
cc55af41e4fc6e2f488c87ae84c8c8c7c0d2275fff41113cca285d716915cc1a	[ "arxiv", "semantic_scholar" ]	Streaming Communication in Multi-Agent Reasoning	Multi-agent reasoning systems adopt a "generate-then-transfer" paradigm that forces end-to-end latency to scale linearly with pipeline depth. We introduce StreamMA, a multi-agent reasoning system that streams each reasoning step to downstream agents as soon as it is generated, pipelining adjacent agents and thus reduci...	[ "Zhen Yang", "Xiaogang Xu", "Wen Wang", "Cong Chen", "Xander Xu", "Ying-Cong Chen" ]	[ "cs.CL", "cs.AI", "cs.MA" ]	[ "Computer Science" ]	2026-06-03T00:00:00	https://arxiv.org/abs/2606.05158	https://arxiv.org/pdf/2606.05158v1	2606.05158	null	0	false	null	null	0
98119941cdd317489c590fd90cf3a17786d8f3890f75e4c487eb501df2c44617	[ "arxiv", "semantic_scholar" ]	GARL: Game-Theoretic Reinforcement Learning for Multi-Agent Strategic Prioritisation	LLM-based multi-agent systems are increasingly used for strategic decision-making tasks. In such settings, performance depends not only on individual model capabilities, but also on the policies by which agents interact and adapt. Multi-agent reinforcement learning can optimise these interaction policies, but its rewar...	[ "Yuxiao Ye", "Yiwen Zhang", "Huiyuan Xie", "Yuqin Huang", "Zhiyuan Liu" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-03T00:00:00	https://arxiv.org/abs/2606.05002	https://arxiv.org/pdf/2606.05002v1	2606.05002	null	0	true	null	null	0
bd7a0013cfd95bda84cea4caa3883cd8619fccf4af5a223a9b33f79fdb4b7351	[ "arxiv", "semantic_scholar" ]	FORGE: Multi-Agent Graduated Exploitation and Detection Engineering	Vulnerability disclosure volumes now far exceed organizational assessment capacity, yet three adjacent research communities (proof-of-concept generation, vulnerability prioritization, and detection rule engineering) operate largely in isolation. Existing automated exploit generation systems report binary pass/fail outc...	[ "Farooq Shaikh" ]	[ "cs.CR", "cs.AI", "cs.MA" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.03453	https://arxiv.org/pdf/2606.03453v1	2606.03453	null	0	false	null	null	0
ed246ae2d4d53d6c2fe32947a8e90375a713eb4d58e84c8b0ca5b9ad0be45988	[ "arxiv", "semantic_scholar" ]	Multi$^2$: Hierarchical Multi-Agent Decision-Making with LLM-Based Agents in Interactive Environments	A central goal of large language model (LLM) research is to build agentic systems that can plan, act, and adapt through sustained interaction with dynamic environments. While recent LLM-based agents exhibit impressive contextual reasoning, their long-horizon decision-making remains fragile, often suffering from objecti...	[ "Sangeun Park", "Minhae Kwon" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.03698	https://arxiv.org/pdf/2606.03698v1	2606.03698	null	0	false	null	null	0
907ca1e93712ec801982a7a54cb468b696ccaabe600e27983715f1d91ed7327d	[ "arxiv", "semantic_scholar" ]	The Deliberative Illusion: Diagnosing Factual Attrition and Stance Homogenization in Multi-Agent LLM Deliberation	Multi-agent LLM systems often treat consensus as evidence of successful interaction. For deliberative problems, however, reliability depends on whether agents preserve the facts and viewpoints needed to interpret an issue. We identify the deliberative illusion: discussion produces (1) factual attrition, the progressive...	[ "Herun Wan", "Jiaying Wu", "Minnan Luo", "Fanxiao Li", "Ningnan Wang", "Nancy F. Chen", "Min-Yen Kan" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.03032	https://arxiv.org/pdf/2606.03032v1	2606.03032	null	0	false	null	null	0
0e49aa46c099be2af4a2920dfe4630d23cda113b2063e1db5abdcc837871915a	[ "arxiv", "semantic_scholar" ]	ToolGate: Token-Efficient Pre-Call Control for Tool-Augmented Vision-Language Agents	Tool-augmented vision-language agents can acquire external perceptual evidence through OCR, detection, segmentation, and other tools, but executing every proposed tool call is costly and sometimes unnecessary. We study the pre-call control problem: after a ReAct-style VLM agent proposes a perceptual tool call, should t...	[ "Anjie Liu", "Yan Song", "Zhixun Chen", "Ziqin Gong", "Zhongwei Yu", "Jun Wang" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.03054	https://arxiv.org/pdf/2606.03054v1	2606.03054	null	0	false	null	null	0
d9737ea0f3ebd22bbe26aa31d16a0c1783d99f683984648c4940027ef25ad7d8	[ "arxiv", "semantic_scholar" ]	SMAC-Talk: A Natural Language Extension of the StarCraft Multi-Agent Challenge for Large Language Models	As LLMs become more widely deployed, they are increasingly expected to work alongside other AI agents rather than operating in isolation. Effective coordination in these settings requires agents to communicate, share information and make decisions under uncertainty. We introduce SMAC-Talk, a natural language extension ...	[ "Joel Sol", "Homayoun Najjaran" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.04202	https://arxiv.org/pdf/2606.04202v1	2606.04202	null	0	false	null	null	0
48cc30f772df1d314fdc41e555f1849aff7edeea092fda30080b4a013836b600	[ "arxiv", "semantic_scholar" ]	A Training-Free Mixture-of-Agents Framework for Multi-Document Summarization using LLMs and Knowledge Graphs	Multi-Document Summarization (MDS) plays a critical role in distilling essential information from collections of textual data. Existing approaches often struggle to capture complex inter-document relationships, rely heavily on large amounts of labeled data for supervised training, or exhibit limited generalization acro...	[ "Cuong Vuong Tuan", "Trang Mai Xuan", "Tien-Cuong Nguyen", "Vu-Duc Ngo", "Thien Van Luong" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.03867	https://arxiv.org/pdf/2606.03867v1	2606.03867	null	0	false	null	null	0
214ac628a6bd2a5d5e5a7422b5fa760739a94e0780394d8a2a70d3f79b07f47c	[ "arxiv", "semantic_scholar" ]	MOC: Multi-Order Communication in LLM-based Multi-Agent Systems	Despite the remarkable progress of Large Language Model (LLM) based Multi-Agent Systems, most research focuses on optimizing coordination topology while largely underexploring the equally critical problem: how to transmit and optimize messages among agents effectively? Current communication schemes typically rely on th...	[ "Yao Guan", "Lin Wang", "Zhihu Lu", "Ziyi Wang", "Wenzhu Yan", "Qiang Duan" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.02359	https://arxiv.org/pdf/2606.02359v1	2606.02359	null	0	true	https://github.com/yao-guan/MOC	null	0
a7705dc96fcfae5ffb9dae0c9eed59a5703751f72de4b27bd94cb03fbf2f7ba1	[ "arxiv", "semantic_scholar" ]	Multi-Agent Computer Use	Computer use agents (CUAs) today are primarily deployed as single serial agents. This setup is suboptimal for complex long-horizon tasks that benefit from task decomposition, parallel execution, and consistent re-planning based on new information. In this paper, we argue that we should instead move towards evaluating a...	[ "Jing Yu Koh", "Ruslan Salakhutdinov", "Daniel Fried" ]	[ "cs.MA", "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.01533	https://arxiv.org/pdf/2606.01533v1	2606.01533	null	0	false	null	null	0
eaf5bba48e2db025c65ddd789de2b1cf4f7599e3504533e0851b69b952944cd1	[ "arxiv", "semantic_scholar" ]	POIROT: Interrogating Agents for Failure Detection in Multi-Agent Systems	Orchestrating Large Language Models into Multi-Agent Systems (LLM-MAS) has unlocked remarkable reasoning capabilities, yet emergent failures and hallucinations that resist characterisation block their deployment in safety-critical domains -- a gap made legally untenable by emerging AI regulation. Existing evaluation pa...	[ "Iñaki Dellibarda Varela", "R. Sendra-Arranz", "Pablo Romero-Sorozabal", "J. M. Valverde-García", "Annemarie F. Laudanski", "Álvaro Gutiérrez", "Eduardo Rocon", "Manuel Cebrian" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.02282	https://arxiv.org/pdf/2606.02282v1	2606.02282	null	0	true	null	null	0
eed585a2aabc91de5dcd1d061fc5fbe870902cd626161b3cbcb46d9e567817fd	[ "arxiv", "semantic_scholar" ]	Ghost Tool Calls: Issue-Time Privacy for Speculative Agent Tools	Tool-augmented language agents speculatively issue likely future tool calls to hide latency, but those calls leak inferred user intent to external services before the agent commits to the branch. Every external observer that received the call retains the disclosure after the agent abandons the branch. Timing is the iss...	[ "Bardia Mohammadi", "Lars Klein", "Akhil Arora", "Laurent Bindschaedler" ]	[ "cs.CR", "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.02483	https://arxiv.org/pdf/2606.02483v1	2606.02483	null	0	false	null	null	0
f885bfe045e03564f1f2248f09226dcbe35a0ac3954e293e7f2b5d3e07331e6b	[ "arxiv", "semantic_scholar" ]	Early Diagnosis of Wasted Computation in Multi-Agent LLM Systems via Failure-Aware Observability	Tool-using multi-agent large language model (LLM) systems spend computation through model tokens, tool calls, retries, and code execution before producing an answer. When a run fails, final-answer evaluation reveals the endpoint but usually not the point at which the trajectory stopped making recoverable progress. This...	[ "Xianyou Li", "Weiran Yan", "Yichao Wu", "Penghao Liang", "Mengwei Yuan", "Jianan Liu", "Jing Yang" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-31T00:00:00	https://arxiv.org/abs/2606.01365	https://arxiv.org/pdf/2606.01365v1	2606.01365	null	0	false	null	null	0
60d2d7a637b17cc96b24e10aa6b12b2784d45ab3b1d12ecf02f932e25124535f	[ "arxiv", "semantic_scholar" ]	FinCom: A Financial Multi-Agent Demo with Disagree-or-Commit Deliberation	Multi-agent systems powered by large language models (LLMs) are increasingly used for financial analysis and decision support. However, existing coordination schemes, especially those emphasizing consensus or debate, are vulnerable to sycophancy: agents conform to peer reasoning instead of evidence, leading to prematur...	[ "Chao Peter Yang", "Zixiao Tan", "Kaisen Yao", "Ziyu Zhou", "Eleanor Jiang", "Michael Wu" ]	[ "cs.MA" ]	[ "Computer Science" ]	2026-05-31T00:00:00	https://arxiv.org/abs/2606.00939	https://arxiv.org/pdf/2606.00939v1	2606.00939	null	0	false	null	null	0
1760d30ff7ffe62b392ebbdcfea4b45c9d49edbd5a175fef9769e277b0b5bc10	[ "arxiv", "semantic_scholar" ]	RAG-driven Multi-Agent LLM Framework with Task Decomposition for Beyond 5G Auto-Configuration	While Large Language Models (LLMs) offer a promising path toward intent-driven network management by translating natural language human intents into machine-readable configurations, they often suffer from hallucinations and structural inconsistencies in multi-step and complex tasks. To address these challenges, this pa...	[ "İrşat Emin Sarıdaş", "Onur Salan", "Ali Görçin", "Ibrahim Hokelek", "Hakan Ali Çırpan" ]	[ "eess.SP" ]	[ "Engineering" ]	2026-05-31T00:00:00	https://arxiv.org/abs/2606.01222	https://arxiv.org/pdf/2606.01222v1	2606.01222	null	0	false	null	null	0
061de5b4a7532b3ac5a8d07f2495caa1d2e1e6b6c3b1269b02daed603ffbdbe3	[ "arxiv", "semantic_scholar" ]	CAREAgent: Clinical Agent with Structured Reasoning and Tool-Integrated for Order Generation	Clinical order generation serves as a critical bridge between clinical decision-making and real-world practice, translating medical decisions into concrete and executable orders. Existing agents mainly focus on coarse-grained decisions and overlook the fine-grained, executable information required for clinical orders. ...	[ "Ruihui Hou", "Ziyue Huai", "Chennuo Zhang", "Ziyan Liu", "Siran Zhao", "Yao Yu", "Jie Zhai", "Tong Ruan" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-31T00:00:00	https://arxiv.org/abs/2606.01094	https://arxiv.org/pdf/2606.01094v1	2606.01094	null	0	false	null	null	0
2e28d4c8ab8557a63792fc114dfd009bbbe192dd6abe2913f6d55905120cf886	[ "arxiv", "semantic_scholar" ]	Scaling Behavior of Single LLM-Driven Multi-Agent Systems	The burgeoning field of LLM-based Multi-Agent Systems (MAS) promises to tackle complex tasks through collaborative intelligence, yet fundamental questions regarding their scaling behavior and intrinsic collective dynamics remain underexplored. This paper systematically investigates how the performance of a homogeneous ...	[ "Jialing Li", "Zhouhong Gu", "Yin Cai", "Hongwei Feng" ]	[ "cs.MA", "cs.AI", "cs.CY" ]	[ "Computer Science" ]	2026-05-30T00:00:00	https://arxiv.org/abs/2606.00655	https://arxiv.org/pdf/2606.00655v1	2606.00655	null	0	false	null	null	0
033102c6d4bde68537276e29358d9dba384232f01d9aba0d22d754713f693ccd	[ "arxiv", "semantic_scholar" ]	FALAT: Tracing Failures in LLM Agent Trajectories via Dependency-Guided Search	LLM-based agents increasingly solve complex tasks through long trajectories involving reasoning steps, tool calls, and inter-agent communication. However, when these agents fail, it is often unclear which agent caused the failure and which step introduced the decisive error. This attribution problem is challenging beca...	[ "Md Nakhla Rafi", "Md Ahasanuzzaman", "Dong Jae Kim", "Zhijie Wang", "Tse-Hsun Chen" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-30T00:00:00	https://arxiv.org/abs/2606.00765	https://arxiv.org/pdf/2606.00765v1	2606.00765	null	0	false	null	null	0
7b5bf44305ea75c39b6a396737e311e99fa30b92251fc32e114e68b2d9a10252	[ "arxiv", "semantic_scholar" ]	Doing What They Say, Not What They Reason: Locating the Faithfulness Gap in LLM Agents	Do LLM agents act on the reasoning they state? This question of process fidelity is central to using LLMs in social simulation, yet it is hard to measure where no reference for correct behavior exists. We study it in acontrolled setting, a Texas Poker simulator with a verifiable reference action for every decision by d...	[ "Yufeng Wang" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-30T00:00:00	https://arxiv.org/abs/2606.00476	https://arxiv.org/pdf/2606.00476v1	2606.00476	null	0	false	null	null	0
24bdd4d7e33f82b77dbc25f96a02b10567101a67206ae0ef6b44dbfcf142433d	[ "arxiv", "semantic_scholar" ]	Not All Flips Are Conformity: Decomposing Stance Convergence in Multi-Agent LLM Debate	Multi-agent debate (MAD) is a promising strategy for improving LLM reasoning, but when agents converge on a shared answer, it is unclear whether that convergence reflects genuine deliberation or social compliance. We show that the conventional answer flip rate conflates three distinct mechanisms: spontaneous instabilit...	[ "Xiqi Hao", "Zengqing Wu", "Yu-Xuan Qiu", "Chuan Xiao", "Ruiqi Xu", "Shuyuan Zheng", "Jianbin Qin" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-30T00:00:00	https://arxiv.org/abs/2606.00820	https://arxiv.org/pdf/2606.00820v1	2606.00820	null	0	false	null	null	0
29e365ea4af0526725d00b7dca82347be015bc98a8f07f08b56fd9f78af1c23a	[ "arxiv", "semantic_scholar" ]	MAVEN: Improving Generalization in Agentic Tool Calling	Generalization across agentic tool-calling environments remains a central challenge for reliable agentic reasoning systems. Although large language models achieve strong results on individual benchmarks, their ability to compose reasoning strategies, preserve intermediate states, and coordinate tools across domains rem...	[ "Omkar Ghugarkar", "Vishvesh Bhat", "Muhammad Ahmed Mohsin", "Asad Aali" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-29T00:00:00	https://arxiv.org/abs/2605.30738	https://arxiv.org/pdf/2605.30738v1	2605.30738	null	0	false	null	null	0
9f12b0c27effb05084be4a97206072e5091f45b9b3f3c724a30b83aa59625311	[ "arxiv", "semantic_scholar" ]	Depth-Dependent Indirect Prompt Injection in Tool-Calling ReAct Agents: Injection Depth, Payload Framing, and Turn-Budget Sensitivity	ReAct agents that interleave chain-of-thought reasoning with tool calls are increasingly deployed for real tasks such as scheduling, file retrieval, and data access. Their tool observation loop creates a direct attack surface: an adversary who controls any tool's return value can embed instructions that redirect the ag...	[ "Mohammadreza Rashidi" ]	[ "cs.CR", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-05-29T00:00:00	https://arxiv.org/abs/2605.30686	https://arxiv.org/pdf/2605.30686v1	2605.30686	null	0	false	null	null	0
0dd89d0eb24c2486f19864db399008f4631087bf819aa48e742671ee9000ee61	[ "arxiv", "semantic_scholar" ]	Counterfactual Graph for Multi-Agent LLM Calibration	Multi-agent LLM systems often treat agreement as evidence: when many agents in a panel give the same answer, that answer is assumed to be more reliable. We show that this assumption can fail after agents communicate. Communication can induce correlated failures and false consensus, so the same vote share may reflect re...	[ "Jiatan Huang", "Mingchen Li", "Ziming Li", "Sunjae Kwon", "Hong Yu", "Chuxu Zhang" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.30653	https://arxiv.org/pdf/2605.30653v1	2605.30653	null	0	false	null	null	0
75b0962d1ccb4d37b3c8f9c842e7ac691eddeef03446ccacd50a93d1c684e61b	[ "arxiv", "semantic_scholar" ]	CONCAT: Consensus- and Confidence-Driven Ad Hoc Teaming for Efficient LLM-Based Multi-Agent Systems	Although large language model (LLM) based multi-agent systems (MAS) show their capability to solve complex tasks and achieve higher performance over single agent systems, they lead to huge computational overheads because of heavy communication between agents. Previous research has made efforts to train a sparse multi-a...	[ "Ziyang Ma", "Dingyi Zhang", "Sichu Liang", "Jiajia Chu", "Pengfei Xia", "Hui Zang", "Deyu Zhou" ]	[ "cs.MA", "cs.CL" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.29612	https://arxiv.org/pdf/2605.29612v1	2605.29612	null	0	false	null	null	0
981d9c2151d336d5b99823916f954bc5486d7b2651716f2fcb8981b3aa7595e3	[ "arxiv", "semantic_scholar" ]	PatchBoard: Schema-Grounded State Mutation for Reliable and Auditable LLM Multi-Agent Collaboration	LLM multi-agent systems often coordinate through natural-language dialogue or loosely structured shared memory, making intermediate state difficult to validate, attribute, and audit. We introduce PatchBoard, a schema-grounded collaboration architecture that replaces inter-agent dialogue with validated JSON Patch mutati...	[ "Shuyu Zhang", "Yaqi Shi", "Lu Wang" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.29313	https://arxiv.org/pdf/2605.29313v1	2605.29313	null	0	false	null	null	0
388efdadf696966dac598bdec8d2e4596750ad6c8bb875561a1903d8e2f07bd0	[ "arxiv", "semantic_scholar" ]	On Effectiveness and Efficiency of Agentic Tool-calling and RL Training	Tool-calling is a central component of modern large language model (LLM) agents, equipping them with skills beyond their parametric knowledge. This paper studies tool-calling along two complementary axes: effectiveness, i.e., how this capability is measured, and efficiency, i.e., how it is learned. On effectiveness, we...	[ "Tong Liu", "Cheng Qian", "Matej Cief", "Yuan He", "Daniele Dan", "Nikolaos Aletras", "Gabriella Kazai" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2606.00135	https://arxiv.org/pdf/2606.00135v1	2606.00135	null	0	false	null	null	0
afdf544749a8553785449ac0db79f315c6451c458ab87a5478416230ef1eee28	[ "arxiv", "semantic_scholar" ]	SkillBrew: Multi-Objective Curation of Skill Banks for LLM Agents	Retrieval-augmented LLM agents increasingly rely on curated skill banks: collections of reusable textual principles that guide decision making on complex tasks. Existing approaches typically expand these banks in an append-only fashion, continuously adding new skills without removing redundant, outdated, or harmful one...	[ "Wentao Hu", "Zhendong Chu", "Yiming Zhang", "Junda Wu", "Ming Jin", "Xiangyu Zhao", "Yilei Shao", "Yanfeng Wang", "Qingsong Wen" ]	[ "cs.CL", "cs.AI", "cs.IR" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.29440	https://arxiv.org/pdf/2605.29440v1	2605.29440	null	0	false	null	null	0
6260d2d569c598ccb827351078219b97204e837de346f7ba606d6b4c38aafe49	[ "arxiv", "semantic_scholar" ]	AgentSchool: An LLM-Powered Multi-Agent Simulation for Education	Despite the rapid deployment of LLMs into classrooms, validating educational AI remains uniquely intractable: interventions act on developing learners whose cognitive and social trajectories are irreversibly shaped, while real-world trials are slow, ethically constrained, and institutionally locked. LLM-based education...	[ "Yulei Ye", "Wenhao Li", "Zhong Wen", "Yunshu Huang", "Yichen Hu", "Zifan Wei", "Yige Wang", "Xinyu Xie", "Haoxuan Yang", "Yanjun Huang", "Ruijia Li", "Hong Qian", "Yu Song", "Bo Jiang", "Bingdong Li", "Lijun Li", "Bo Zhang", "Pinlong Cai", "Xingcheng Xu", "Shuangye Chen", "X...	[ "cs.AI", "cs.MA" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.30144	https://arxiv.org/pdf/2605.30144v1	2605.30144	null	0	false	null	null	0
25e6120dfe0689e19541713972fbd77c03487c86afc0b754d3f7b4c3ee0b60b3	[ "arxiv", "semantic_scholar" ]	MINDGAMES: A Live Arena for Evaluating Social and Strategic Reasoning in Multi-Agent LLMs	Large language models (LLMs) are increasingly deployed as interactive agents, yet their capacity for social and strategic reasoning over extended interaction remains poorly understood. Existing evaluations rely on static vignettes or single-game benchmarks that cannot capture the sustained, multi-faceted reasoning that...	[ "Kevin Wang", "Anna Thöni", "Benjamin Kempinski", "Bobby Cheng", "Jianzhu Yao", "Benjamin Finch", "Leon Guertler", "Viraj Nadkarni", "Yihan Jiang", "Aliaksei Korshuk", "Alexander Buyantuev", "Ilya Makarov", "Siyuan Wu", "Yu-Chi Cheng", "Yan-Ru Ju", "Ti-Rong Wu", "I-Hsuan Chu", "Yu-...	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.29512	https://arxiv.org/pdf/2605.29512v1	2605.29512	null	0	false	null	null	0
cb4c683408ec46a64be6f16bc532595df67485a27bffd95a1311307e1972bcf2	[ "arxiv", "semantic_scholar" ]	Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems	LLM-based multi-agent systems (MAS) have emerged as an effective paradigm for complex and long-horizon tasks. However, in real-world tasks, MAS often exhibit various failures during execution and such failures are difficult to eliminate during design. This motivates experience-driven MAS evolution, where a system impro...	[ "Zhezheng Hao", "Tianfu Wang", "Huanshuo Dong", "Ziyan Liu", "Hong Wang", "Xiankun Lin", "Qiang Lin", "Can Wang", "Hande Dong", "Jiawei Chen" ]	[ "cs.MA", "cs.AI" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.29790	https://arxiv.org/pdf/2605.29790v1	2605.29790	null	0	false	null	null	0
db45cb0fa85463afa95df3aa5bce3c6ad65e7b66a530d99290d8a5b260772206	[ "arxiv", "semantic_scholar" ]	LLM-ALSO: LLM-Driven Adaptive Learning-Signal Optimization for Multi-Agent Reinforcement Learning	Effective training-time guidance is central to multi-agent reinforcement learning (MARL), yet remains difficult in sparse-reward settings where weak supervision limits coordination and policy improvement, and existing methods often require substantial domain expertise or manual design effort. Large language models (LLM...	[ "Xiaoguang Wu", "Zhi Zheng", "Hui Xiong" ]	[ "cs.MA" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.29293	https://arxiv.org/pdf/2605.29293v1	2605.29293	null	0	false	null	null	0
3a40ccc0e6b9f8facf36fed31e50335bbb981e646c48133ca2f5c263a2a910c3	[ "arxiv", "semantic_scholar" ]	A Theory-Guided LLM Pedagogical Agent for STEM+C Scaffolding Without Over-Reliance	LLM pedagogical agents are proliferating, yet recent findings have raised questions about their adherence to established theories of learning and, by extension, their educational value. Concerns regarding cognitive offloading, over-reliance, and "gaming" behaviors persist and remain largely unaddressed. In response, we...	[ "Clayton Cohn", "Surya Rayala", "Siyuan Guo", "Hanchen David Wang", "Naveeduddin Mohammed", "Umesh Timalsina", "Shruti Jain", "Ryan Li", "Angela Eeds", "Menton Deweese", "Pamela J. Osborn Popp", "Rebekah Stanton", "Shakeera Walker", "Ashwin T S", "Meiyi Ma", "Gautam Biswas" ]	[ "cs.MA" ]	[ "Computer Science" ]	2026-05-28T00:00:00	https://arxiv.org/abs/2605.30539	https://arxiv.org/pdf/2605.30539v1	2605.30539	null	0	false	null	null	0
5f260cd750da3734ae9c16ffc723e7a50dcead9c1b10cb84b29ec0d91e4f1b14	[ "arxiv", "semantic_scholar" ]	MRMMIA: Membership Inference Attacks on Memory in Chat Agents	Membership inference attacks (MIAs) test whether a target data record belongs to a system's private data, and have become a standard tool to measure privacy leakage in machine learning systems. Prior work has primarily focused on training corpora or retrieval databases. However, MIAs against agent memory have received ...	[ "Kai Chen", "Yan Pang", "Tianhao Wang" ]	[ "cs.CR", "cs.LG" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.27825	https://arxiv.org/pdf/2605.27825v1	2605.27825	null	0	false	null	null	0
b571632241f894b466e0d6089e77d9ae1a452b0b05f647355a807d3bce9b0047	[ "arxiv", "semantic_scholar" ]	Do Agents Know What They Can't Do? Evaluating Feasibility Awareness in Tool-Using Agents	Tool-using agents often incur substantial computational cost due to long reasoning chains and iterative tool usage. In practical scenarios, many tasks become infeasible under constrained tool environments, where the capabilities required for successful task completion are unavailable. Detecting infeasible tasks and sto...	[ "Liang Cheng", "Mingsheng Cai", "Jiuming Jiang", "Luo Mai" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.28532	https://arxiv.org/pdf/2605.28532v1	2605.28532	null	0	false	null	null	0
3e990cb9fa5d46becb63ac9d1ebd22a1e8efb800e3b6826a95a743f7c1971199	[ "arxiv", "semantic_scholar" ]	When Does Memory Help Multi-Trajectory Inference for Tool-Use LLM Agents?	Multi-trajectory inference for tool-use LLM agents - generating multiple reasoning attempts and selecting among them - benefits from transferring knowledge across attempts so that later ones avoid the pitfalls of earlier ones. Existing cross-trajectory memory methods (trajectory-level reflection, atomic fact extraction...	[ "Xinzhe Li", "Yaguang Tao" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.28224	https://arxiv.org/pdf/2605.28224v1	2605.28224	null	0	false	null	null	0
298af24472dbb0fe6598ee22abb00ccb1d2753a190447d54fb40b42c383ccd99	[ "arxiv", "semantic_scholar" ]	EgoBench: An Interactive Egocentric Multimodal Benchmark for Tool-Using Agents	As AI agents increasingly operate in open, real-world environments, they require a deep synergy of multimodal perception, tool invocation with multi-hop reasoning, and dynamic interaction with users. However, existing benchmarks fail to jointly evaluate these capabilities due to challenges in designing strictly coupled...	[ "Yunqi Liu", "Tong Niu", "Zitong Wang", "Zhenlong Dai", "Yuqi Qing", "Weiqiang Wang", "Jian Liu" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.27820	https://arxiv.org/pdf/2605.27820v1	2605.27820	null	0	false	null	null	0
87fbbf227c0400982934dcb3daceafc431a7097dec19cf5896d8d8bc65f093ca	[ "arxiv", "semantic_scholar" ]	A Unified Framework for the Evaluation of LLM Agentic Capabilities	As LLMs are increasingly deployed as agents, reliable assessment of their agentic capabilities has become essential. However, reported benchmark scores often jointly reflect model capability and the implementation choices each benchmark is packaged with, making cross-benchmark results difficult to interpret as clean me...	[ "Pengyu Zhu", "Lijun Li", "Yaxing Lyu", "Qianxin Luo", "Jingyi Yang", "Yi Liu", "Tingfeng Hui", "Xinyu Yuan", "Li Sun", "Sen Su", "Jing Shao" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.27898	https://arxiv.org/pdf/2605.27898v1	2605.27898	null	0	true	https://github.com/whfeLingYu/A-Unified-Framework-for-the-Evaluation-of-LLM-Agentic-Capabilities	null	0
636a5875908c40b43dcc168f16de097b2061f80c516abf87589ce92f0a427cef	[ "arxiv", "semantic_scholar" ]	Tool Forge: A Validation-Carrying Toolchain for Governed Agentic Execution	Large language model agents are increasingly expected to perform operational work: calling APIs, manipulating files, assembling workflows, and acting inside enterprise systems. Yet the tool layer on which this execution depends is still commonly treated as either a hand-written integration artifact or a static list of ...	[ "Swanand Rao" ]	[ "cs.SE", "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.28000	https://arxiv.org/pdf/2605.28000v1	2605.28000	null	0	true	https://github.com/nextmoca/tool-forge	null	0
f167587e9632604783758250afc8647bf748581dbd6a4b0835d38dcc4f9cf363	[ "arxiv", "semantic_scholar" ]	MolLingo: Molecule-Native Representations for LLM-Powered Scientific Agents	We present MolLingo, a multi-agent system that emulates the reasoning process of a chemist to automate molecular design. Existing LLM-based approaches either operate as standalone generative models without access to external tools or lack the multi-agent coordination and shared memory needed for iterative, evidence-dri...	[ "Thao Nguyen", "Heng Ji" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.27853	https://arxiv.org/pdf/2605.27853v1	2605.27853	null	0	true	null	null	0
3b6d4f6906421ba85a20786b5742c4096ef3c6199d4dd9742d6becc153e2774d	[ "arxiv", "semantic_scholar" ]	Mixture-of-Experts Knowledge Graph Retrieval-Augmented Generation for Multi-Agent LLM-based Recommendation	Large language models (LLMs) have recently been adopted for recommendations due to their ability to understand user intent and item semantics. However, LLM-based recommender systems often rely on parametric knowledge and suffer from outdated knowledge, motivating knowledge graph retrieval-augmented generation (KG-RAG) ...	[ "Shijie Wang", "Chengyi Liu", "Yujuan Ding", "Shanru Lin", "See-Kiong Ng", "Xu Xin", "Wenqi Fan" ]	[ "cs.IR" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.28175	https://arxiv.org/pdf/2605.28175v2	2605.28175	10.1145/3770855.3817630	0	false	null	null	0
efced1838dcbca799f62038b63df0f0176a1d35e280bd3862d9599af81a98441	[ "arxiv", "semantic_scholar" ]	Examining Agents' Bias Amplification versus Suppression in Multi-Agent Systems	Multi-agent systems are increasingly deployed to support various tasks where agents interact to achieve individual and collective objectives. Although these systems can enhance task performance and decision-making, fairness preservation through bias reduction remains challenging. This study examines how agent-level bia...	[ "Zejian Eric Wu", "Zhongyi Jiang", "Yuan Zhuang", "Paul Jen-Hwa Hu" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.28098	https://arxiv.org/pdf/2605.28098v1	2605.28098	null	0	false	null	null	0
04e912a4a1454289103316cb797db2a6221471613207c2397c53a83bdbae5ce6	[ "arxiv", "semantic_scholar" ]	Multi-Agent LLM-based Metamorphic Testing for REST APIs	As REST APIs become an increasingly significant part of software systems, their validation is becoming more critical. Hence, testing and uncovering underlying issues are of utmost importance for improving software quality. However, testing REST APIs is challenging mainly due to the difficulty of assessing whether the o...	[ "Shehroz Khan", "Abdullah Mughees", "Gaadha Sudheerbabu", "Tanwir Ahmad", "Dragos Truscan" ]	[ "cs.SE", "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.28321	https://arxiv.org/pdf/2605.28321v1	2605.28321	null	0	false	null	null	0
6f8f94bfe504b805588edd99590da35f79a7648c0bd603ad55134114eb80f932	[ "arxiv", "semantic_scholar" ]	SafeRx-Agent: A Knowledge-Grounded Multi-Agent Framework for Safe and Explainable Medication Recommendation	Medication recommendation predicts medications for patient visits, but existing methods still face two key challenges. At the model level, traditional drug recommendation methods only predict structured drug codes with limited evidence grounding, while LLM agents can use richer clinical context but may lack safety veri...	[ "Xinyu Wang", "Hanwei Wu", "Zhenghan Tai", "Sicheng Lyu", "Qincheng Lu", "Ziyu Zhao", "Jijun Chi", "Jingrui Tian", "Xiao-Wen Chang", "Ziyang Song" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.29146	https://arxiv.org/pdf/2605.29146v2	2605.29146	null	0	false	null	null	0
fede12f8835ea170c097626204cf8028ee4cca81e490000c0238c1f9322d2578	[ "arxiv", "semantic_scholar" ]	Voluntary Collusion with Secret Tools in Competing LLM Agents	Even when a tool is explicitly described as unfair and harmful to others, ostensibly safety-aligned LLM agents still voluntarily engage in secret collusion whenever doing so confers a strategic advantage. To investigate this phenomenon, we introduce an empirical framework built on two strategic multi-agent environments...	[ "Xijie Zeng", "Frank Rudzicz" ]	[ "cs.AI", "cs.MA" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27593	https://arxiv.org/pdf/2605.27593v1	2605.27593	null	0	false	null	null	0
0d203e77f5144e6b09082f114c73a939208e365fdb42e0029ad146815f6976d6	[ "arxiv", "semantic_scholar" ]	Agents that Matter: Optimizing Multi-Agent LLMs via Removal-Based Attribution	As multi-agent systems (MAS) become increasingly complex, identifying the contributions of individual agents is critical for system optimization. However, existing approaches lack a rigorous, unified framework for credit assignment. In this work, we formalize agent attribution as a cooperative game, parameterized by th...	[ "Mingyu Lu", "Yushan Huang", "Chris Lin", "Su-In Lee" ]	[ "cs.MA", "cs.CL" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27621	https://arxiv.org/pdf/2605.27621v1	2605.27621	null	0	false	null	null	0
620fbe3ccab184dfebd8b7248207d8e447810dee1bbad7a5984359368f8da6a8	[ "arxiv", "semantic_scholar" ]	HARP: Measuring Harm Amplification in Multi-Agent LLM Systems	Multi-agent LLM systems decompose workflows across agents, tools, shared context, memory, and decision gates. This modularity improves interpretability, but creates a propagation risk: a bounded perturbation to one component can be reused by other agents and amplified into system-level harm. We introduce HARP (Harm Amp...	[ "Md Hafizur Rahman", "Zafaryab Haider", "Tanzim Mahfuz", "Prabuddha Chakraborty" ]	[ "cs.CR", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27489	https://arxiv.org/pdf/2605.27489v1	2605.27489	null	0	false	null	null	0
53421d40fd629663042ce04bd89500470a939e5bc52350ef86f09993fa899067	[ "arxiv", "semantic_scholar" ]	Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent Systems	LLM safety evaluations predominantly test models in isolation, yet deployed AI agents increasingly operate within persistent social environments alongside other agents. We introduce a Moltbook-style simulation platform where thousands of LLM agents interact across communities over a simulated month, and use it to evalu...	[ "Aman Priyanshu", "Supriti Vijay", "Esha Pahwa" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27766	https://arxiv.org/pdf/2605.27766v1	2605.27766	null	0	false	null	null	0
ecb6d140e8b7f9bbf5d830b1469fb0badac665c5ef2e425fa9d3d0dee437ad6e	[ "arxiv", "semantic_scholar" ]	UnityMAS-O: A General RL Optimization Framework for LLM-Based Multi-Agent Systems	LLM-based multi-agent systems decompose complex tasks into interacting roles, but most remain manually orchestrated by prompts, tools, and control rules, while agents are rarely optimized through a unified reinforcement learning interface. Existing RL post-training frameworks mainly target single-policy optimization an...	[ "Yiqun Chen", "Wei Yang", "Erhan Zhang", "Shijie Wang", "Qi Liu", "Zechun Niu", "Bin Zhang", "Haitao Li", "Rui Li", "Lingyong Yan", "Jinyuan Feng", "Biqing Qi", "Xiaochi Wei", "Yan Gao", "Yi Wu", "Yao Hu", "Jiaxin Mao" ]	[ "cs.AI", "cs.CL", "cs.MA" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.26646	https://arxiv.org/pdf/2605.26646v1	2605.26646	null	0	false	null	null	0
dea4308531e83625260e5845be78765ffa9e4b722b0617582a38b7c87bc19a39	[ "arxiv", "semantic_scholar" ]	TADDLE: A Tool-Augmented Agent for Detecting Deficient LLM-Generated Peer Reviews	LLM-generated peer reviews are increasingly common at major venues, yet their deficiencies are hard to detect because they are uniformly fluent and well-structured. Existing work either classifies authorship without judging quality, or scores quality with features designed for human-written reviews; no prior system det...	[ "Hanqi Duan", "Xiang Li" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.26911	https://arxiv.org/pdf/2605.26911v1	2605.26911	null	0	true	https://github.com/AquariusAQ/TADDLE	null	0
2a3ff62c9f2e7fc34a7272db86f0a5a971de85a29b07cc8bad4c812e3012ceae	[ "arxiv", "semantic_scholar" ]	Decoupled Intelligence: A Multi-Agent LLM Framework for Controllable Traffic Scenario Generation in SUMO	The integration of Large Language Models (LLMs) with microscopic traffic simulation offers a promising path toward autonomous urban planning and intelligent transportation analysis. However, existing monolithic agent architectures often struggle with the complexity of end-to-end simulation workflows, leading to reasoni...	[ "Shuyang Li", "Ruimin Ke" ]	[ "cs.MA", "cs.HC" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27685	https://arxiv.org/pdf/2605.27685v1	2605.27685	null	0	false	null	null	0
2c7ad6f14d755a8643947ad70d3a8d5b654ac7b7d6b02bcbe6785b5e11107049	[ "arxiv", "semantic_scholar" ]	FinHarness: An Inline Lifecycle Safety Harness for Finance LLM Agents	Finance LLM agents must simultaneously block prompt-induced unauthorized actions and approve legitimate multi-step business workflows. However, boundary filters often miss irreversible mid-trajectory tool calls, while post-hoc LLM judges perform auditing only after termination -- too late for intervention and at a comp...	[ "Haoxuan Jia", "Yang Liu", "Bin Chong", "Yingguang Yang", "Yancheng Chen", "Jiayu Liang", "Qian Li", "Hanning Lu", "Kefu Xu", "Hao Zheng", "Chongyang Zhang", "Hao Peng", "Philip S. Yu" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27333	https://arxiv.org/pdf/2605.27333v1	2605.27333	null	0	false	null	null	0
2d654f81bf2e8c5c69a936a1961a9776cc8db723783f4e753a29cedfa12deb11	[ "arxiv", "semantic_scholar" ]	TRACES: Proactive Safety Auditing for Multi-Turn LLM Agents via Trajectory-State Modeling	LLM agents increasingly operate through multi-turn tool use and environment interaction, where safety risks often emerge from intermediate steps long before they surface in the final outcome. Reactive auditing is therefore insufficient: post-hoc diagnosis frequently misses the chance to flag risks while they are unfold...	[ "Jiaqian Li", "Yanshu Li", "Boxuan Zhang", "Ruixiang Tang", "Kuan-Hao Huang" ]	[ "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27690	https://arxiv.org/pdf/2605.27690v1	2605.27690	null	0	false	null	null	0
0d45dfe7d74798c43d6b16da136b221aa25b085e1e3461071a9473f3f3062cef	[ "arxiv", "semantic_scholar" ]	A Policy-Driven Runtime Layer for Agentic LLM Serving	Multi-agent LLM systems have become the dominant production workload, but the serving stack was not built for them. The agent framework above knows agent identities, role, schemas, and dispatch structure but never sees an engine-level event; the serving engine below sees every event but knows nothing about agents. A su...	[ "Rui Zhang", "Chaeeun Kim", "Liting Hu" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.27744	https://arxiv.org/pdf/2605.27744v1	2605.27744	null	0	false	null	null	0
7096f71493f2f15e0247dcba31891b8d20dee6aaa2ac549d1f7517d00f1d5e62	[ "arxiv", "semantic_scholar" ]	Helicase: Uncertainty-Guided Supply Chain Knowledge Graph Construction with Autonomous Multi-Agent LLMs	LLM-based multi-agent systems have been widely adopted for knowledge retrieval and report generation, synthesizing known information through web search and textual reasoning. However, many critical information tasks in supply chains are not simple one-shot queries: they are structural inference problems requiring multi...	[ "Yunbo Long", "Haolang Zhao", "Ge Zheng", "Alexandra Brintrup" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-26T00:00:00	https://arxiv.org/abs/2605.26835	https://arxiv.org/pdf/2605.26835v1	2605.26835	null	0	false	null	null	0
7dd6a05ef0aa25e73f9ff50d484c2dbc9fee6a8da8e026445ef5f96408059b0c	[ "arxiv", "semantic_scholar" ]	Stateful Inference for Low-Latency Multi-Agent Tool Calling	Multi-agent tool calling is becoming the dominant interaction pattern for LLM-based systems, yet existing inference frameworks treat each tool call as an independent request, re-processing the entire conversation from scratch even though 85-95% of the prompt is unchanged from the previous turn. We present a stateful in...	[ "Victor Norgren" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-05-25T00:00:00	https://arxiv.org/abs/2605.26289	https://arxiv.org/pdf/2605.26289v1	2605.26289	null	0	false	null	null	0
d01e686ac0d679740def537aa7e2fda81620f9155ba80fa4a0d331c0fb9fff13	[ "arxiv", "semantic_scholar" ]	Tool-Call Dependency Structure is Linearly Decodable in LLM Agent Residual Streams	Tool-using LLM agents produce trajectories whose calls form a directed dependency graph: earlier tool outputs supply arguments to later calls. Whether this execution structure is represented inside the model is unknown; prior structural probes have targeted static code or chain-of-thought text, not an agent's run-time ...	[ "Tianda Sun", "Dimitar Kazakov" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-25T00:00:00	https://arxiv.org/abs/2605.25310	https://arxiv.org/pdf/2605.25310v1	2605.25310	null	0	false	null	null	0
27fc16535a85de67fd41f03476c888ac520b2b5ed071f1db8ff78607ec761cd1	[ "arxiv", "semantic_scholar" ]	Multi-Agent Systems are Mixtures of Experts: Who Becomes an Influencer?	The effectiveness of multi-agent LLM deliberation depends not only on the agents' individual predictions, but also on how they communicate and collaborate. We study this mechanism through the lens of Friedkin-Johnsen (FJ) opinion dynamics, a tractable model for analyzing stubbornness, influence, and opinion change in m...	[ "Franka Bause", "Jonas Niederle", "Martin Pawelczyk", "Rebekka Burkholz" ]	[ "cs.MA", "cs.LG" ]	[ "Computer Science" ]	2026-05-25T00:00:00	https://arxiv.org/abs/2605.25929	https://arxiv.org/pdf/2605.25929v1	2605.25929	null	0	false	null	null	0
5572df65022f8c3c826324972a5e2a769bc24d5a866917f2ec6a00c4c2784fe7	[ "arxiv", "semantic_scholar" ]	DarkForest: Less Talk, Higher Accuracy for Multi-Agent LLMs	Multi-agent LLM systems improve reasoning by combining outputs from multiple agents, but interaction-heavy methods can introduce error propagation and high communication overhead. When agents exchange raw responses or reasoning traces, incorrect intermediate reasoning may be adopted and amplified, leading to confident ...	[ "Yi Li", "Songtao Wei", "Dongming Jiang", "Zhichun Guo", "Qiannan Li", "Bingzhe Li" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-24T00:00:00	https://arxiv.org/abs/2605.25188	https://arxiv.org/pdf/2605.25188v1	2605.25188	null	0	false	null	null	0
e0d07677e75375282ab6a07559010d907f4238b10a8fc8a775506179a9f0c9e9	[ "arxiv", "semantic_scholar" ]	Meta-Agent: From Task Descriptions to Verified Multi-Agent Systems	AI agents are increasingly used to solve complex, multi-step tasks, but existing multi-agent frameworks remain brittle as workflows grow in scale and depth. Small errors at intermediate stages can propagate through agent interactions, while insufficient grounding and weak verification mechanisms further limit reliabili...	[ "Andy Xu", "Yu-Wing Tai" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-24T00:00:00	https://arxiv.org/abs/2605.25233	https://arxiv.org/pdf/2605.25233v1	2605.25233	null	0	false	null	null	0
49e1b39966d635dfb24243186f944331b601c28113aa494cc84a046df91c9732	[ "arxiv", "semantic_scholar" ]	GroupTravelBench: Benchmarking LLM Agents on Multi-Person Travel Planning	Travel planning in the real world is overwhelmingly a \textit{group} activity, yet existing LLM travel-planning benchmarks reduce it to a single user, where the field is approaching saturation. This single-user assumption sidesteps what makes group planning hard for an agent: discovering private preferences across mult...	[ "Xiang Cheng", "Yulan Hu", "Lulu Zheng", "Zheng Pan", "Xin Li", "Yong Liu" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-24T00:00:00	https://arxiv.org/abs/2605.25200	https://arxiv.org/pdf/2605.25200v2	2605.25200	null	1	false	null	null	0.0753
1dbf3d581ffd95e0056adc109b507c93b1468be88bddaf49e60d13127e5a8d69	[ "arxiv", "semantic_scholar" ]	CP-Agent: A Calibrated Risk-Controlled Agent for Feedback-Driven Competitive Programming	Large language models still struggle with contest-level programming, while many agentic remedies rely on massive inference-time sampling or expensive multi-stage post-training. We study when execution feedback reliably helps an LLM CP solver and which mechanisms govern the gains. We model feedback-driven solving as a c...	[ "Peisong Wang", "Bowen Liu", "Zehua Li", "Yuyao Wang", "Zhiwei Ma", "Yuhan Li", "Jia Li" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-23T00:00:00	https://arxiv.org/abs/2605.24693	https://arxiv.org/pdf/2605.24693v1	2605.24693	null	0	true	https://github.com/NineAbyss/CP-Agent	null	0
1f475af873f519e84f6aae61b97fe2366c659ab51ea106281bd2c6547671b58c	[ "arxiv", "semantic_scholar" ]	Market Regime Council for Dynamic Credit Assignment in Multi-Agent LLM Decision Systems	Multi-agent LLM decision systems for portfolio management still lack a principled way to assign credit across specialist agents, remain vulnerable to cold-start dominance under regime shifts, and offer limited transparency into how final allocations are formed. We propose Market Regime Council (MRC), a cooperative mult...	[ "Yunhua Pei", "Zerui Ge", "Jin Zheng", "John Cartlidge" ]	[ "cs.AI", "cs.LG", "q-fin.PM" ]	[ "Computer Science", "Economics" ]	2026-05-23T00:00:00	https://arxiv.org/abs/2605.24490	https://arxiv.org/pdf/2605.24490v1	2605.24490	null	0	false	null	null	0

End of preview. Expand in Data Studio

LLM Agent & Tool-Use Papers — FineSet

Continuously-updated collection of research papers on LLM agents and tool use (function calling, multi-agent, ReAct, AutoGPT), assembled and deduplicated by FineSet from arXiv and Semantic Scholar.

Dataset details

Records: 1660
Date range: 2023–2026
Sources: arXiv, Semantic Scholar (cross-referenced, duplicates merged)
arXiv categories: cs.LG, cs.CL
Quality scoring: citation-normalized, 0–1 (p50=0.15, p90=0.377)
Papers with code: 355
Format: JSONL, one record per line

Fields

Field	Type	Description
id	string	Deterministic SHA256 record id
sources	list	Which sources contributed (`arxiv`, `semantic_scholar`)
title	string	Paper title
abstract	string	Full abstract
authors	list	Author names
categories	list	arXiv category codes
fields_of_study	list	Semantic Scholar field tags
published_date	string	ISO 8601 date
url	string	arXiv abstract URL
pdf_url	string\|null	Open-access PDF if available
arxiv_id	string\|null	arXiv identifier
doi	string\|null	DOI if available
citation_count	int	Citation count (Semantic Scholar)
influential_citation_count	int	Influential citations (Semantic Scholar)
has_code	bool	Code repo detected in the arXiv comment
code_url	string\|null	GitHub URL if detected
venue	string\|null	Publication venue
quality_score	float	0–1, citation-normalized

Quality score methodology

quality_score = min(1.0, log10(citation_count + 1) / 4)

A citation-normalized heuristic: 0 for uncited papers, ~0.5 at 100 citations, ~0.75 at 1,000, 1.0 at 10,000+. Useful for filtering training data by impact.

Want daily-updated datasets on your research topic?

→ fineset.io — describe what you want to train on, get a dataset. Join the waitlist.

Downloads last month: 8