arXiv:2605.25338v1 Announce Type: cross Abstract: Large language model (LLM) agents frequently fail on multi-step tasks involving reasoning, tool use, and environment interaction. While such failures are typically logged or retried heuristically, they contain structured signals a…
arXiv:2605.23929v1 Announce Type: new Abstract: Modern AI systems increasingly rely on workflows composed of multiple interacting agents, some powered by large language models (LLMs) and others by conventional computational modules. This paper analyzes the fundamental tradeoffs b…
arXiv cs.AI
TIER_1English(EN)·Wenqian Ye, Bo Yuan, Zhichao Xu, Yijun Tian, Yawei Wang, Henry Kautz, Aidong Zhang·
arXiv:2605.24197v1 Announce Type: new Abstract: We study a class of emergent misalignment in multi-agent systems (MAS), with a focus on automated workflows, which we refer to agentic misalignment. Although these systems can solve complex tasks, they often fail because agents act …
arXiv:2605.24202v1 Announce Type: new Abstract: Multi-agent LLM workflows route inference through specialized roles to lift end-task accuracy, but jointly training those roles with reinforcement learning is unstable in ways that are poorly understood. We study when end-to-end RL …
arXiv cs.AI
TIER_1English(EN)·Harshada Badave, Santosh Borse, Andrea Gomez, Harshitha Narahari, Sara Carter, Vishwa Bhatt, Aishani Rachakonda, Shuxin Lin, Dhaval Patel·
arXiv:2605.24219v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly deployed as autonomous agents that reason, use tools, and act over multiple steps. Yet most hallucination benchmarks still evaluate only the final output, missing failures that originate…
arXiv:2605.24486v1 Announce Type: new Abstract: Recent progress on long-horizon agentic tasks has been driven largely by scaling up individual agents through stronger models, better tools, and more effective scaffolding. In contrast, much less is understood about scaling out: whe…
arXiv:2605.24598v1 Announce Type: new Abstract: Large language model (LLM) agents excel at solving complex long-horizon tasks through autonomous interaction with environments. However, their real-world deployment faces a fundamental device--cloud dilemma: on-device models are eff…
arXiv cs.AI
TIER_1English(EN)·Zhimin Lin, Kun Cheng, Fan Bai, Jie Gao·
arXiv:2605.24600v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used for qualitative data analysis (QDA), yet their outputs often miss the depth and nuance of human analysis. We argue this gap reflects a missing credibility practice from human QDA: p…
arXiv:2605.24775v1 Announce Type: new Abstract: Operating LLMs as coordinated multi-agent research systems over multi-hour runs surfaces failure modes that single-shot evaluation cannot: upstream providers throttle without warning, sub-agents drift the task to fit accessible tool…
arXiv:2605.24823v1 Announce Type: new Abstract: Manufacturing has passed through four widely recognized paradigms - mechanization, electrification, programmable automation, and Smart Manufacturing - each defined by the kind of work it shifted from humans to machines. In every cas…
arXiv:2605.25188v1 Announce Type: new Abstract: 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 trac…
arXiv:2605.25233v1 Announce Type: new Abstract: 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 interaction…
arXiv:2605.25815v1 Announce Type: new Abstract: Agent-to-Agent (A2A) networks enable autonomous AI agents to collaborate by sharing reusable problem-solving instructions. However, how these decentralized ecosystems operate in practice remains largely unexplored. We present the fi…
arXiv cs.AI
TIER_1English(EN)·Nikos Pagonas, Matthew Lou, Tianyi Peng, Dan Rubenstein, Kostis Kaffes·
arXiv:2605.23914v1 Announce Type: cross Abstract: Agentic workflows interleave configurable LLM stages with tool stages and often include retries or refinement loops. Existing workflow managers profile full workflow configurations offline and assign each request a static workflow…
arXiv:2605.23949v1 Announce Type: cross Abstract: As Large Language Models (LLMs) evolve into interactive agents, understanding their behavioral alignment within human social dynamics becomes essential. While behavioral game theory offers a framework to study these interactions, …
arXiv cs.AI
TIER_1English(EN)·Darek Kleczek, Fuheng Zhao, Alexander W. Lee, Julien Tissier, Pawel Liskowski, Ugur Cetintemel, Anupam Datta·
arXiv:2605.24183v1 Announce Type: cross Abstract: We introduce AvalancheBench, a benchmark for evaluating enterprise data agents through \emph{latent world recovery}. AvalancheBench improves on existing benchmarks in three ways. First, it evaluates analytical understanding rather…
arXiv cs.AI
TIER_1English(EN)·Nesreen K. Ahmed, Nima Nafisi·
arXiv:2605.24216v1 Announce Type: cross Abstract: Monitoring autonomous large language model (LLM) agents for covert malicious behavior is challenging due to delayed, context-dependent, and long-horizon attack patterns. Agents may pursue hidden objectives while maintaining superf…
arXiv:2605.24453v1 Announce Type: cross Abstract: Large Language Model (LLM)-based code analysis tools are adopted to automate software documentation tasks. However, the scalability of these approaches to real codebases, where Intermediate Representations (IR) exceed LLM context …
arXiv:2605.25746v1 Announce Type: cross Abstract: As large language model (LLM)-based multi-agent systems scale to handle increasingly complex tasks, balancing structural stability and dynamic adaptability becomes increasingly challenging. Existing systems typically adopt either …
arXiv:2605.25920v1 Announce Type: cross Abstract: While large language models (LLMs) augmented with agentic search capabilities show promise for legal reasoning, they overlook a fundamental constraint that applicable law must match the temporal context of each case, as retroactiv…
arXiv cs.AI
TIER_1English(EN)·Tatiana Petrova (SEDAN SnT, University of Luxembourg, Luxembourg, Luxembourg), Boris Bliznioukov (SEDAN SnT, University of Luxembourg, Luxembourg, Luxembourg), Aleksandr Puzikov (SEDAN SnT, University of Luxembourg, Luxembourg, Luxembourg), Radu State (S…·
arXiv:2507.10644v4 Announce Type: replace Abstract: The Web of Agents (WoA) transforms the document-centric Web into an environment of autonomous agents acting on users' behalf, a vision newly tractable as large language models (LLMs) mature. We argue that across three decades th…
arXiv:2601.03624v3 Announce Type: replace Abstract: The rapid evolution of Large Language Models (LLM) and subsequent Agentic AI technologies requires systematic architectural guidance for building sophisticated, production-grade systems. This paper presents an approach for archi…
arXiv:2602.03955v3 Announce Type: replace Abstract: While large language model (LLM) multi-agent systems achieve superior reasoning performance through iterative debate, practical deployment is limited by their high computational cost and error propagation. This paper proposes Ag…
arXiv:2603.28716v2 Announce Type: replace Abstract: Agentic RL can benefit substantially from reusable experience, yet existing skill-based methods mainly extract trajectory-level guidance and often lack principled mechanisms for maintaining an evolving skill memory. We propose D…
arXiv:2604.11557v2 Announce Type: replace Abstract: Tool-use capability is a fundamental component of LLM agents, enabling them to interact with external systems through structured function calls. However, existing research exhibits inconsistent interaction representations, large…
arXiv:2605.04906v2 Announce Type: replace Abstract: While Large Language Models (LLMs) excel in certain reasoning tasks, they struggle in multi-agent games where the final outcome depends on the joint strategies of all agents. In multi-agent games, the non-stationarity of other a…
arXiv cs.AI
TIER_1English(EN)·Simon Yu, Derek Chong, Ananjan Nandi, Dilara Soylu, Jiuding Sun, Christopher D Manning, Weiyan Shi·
arXiv:2605.10913v2 Announce Type: replace Abstract: As LLM agent systems take on more complex tasks, they increasingly rely on meta-agents: higher-order agents that operate on other agents, much as managers supervise employees. Whatever a meta-agent does: coordinating agents, hal…
arXiv:2602.03695v2 Announce Type: replace-cross Abstract: While existing multi-agent systems (MAS) can handle complex problems by enabling collaboration among multiple agents, they are often highly task-specific, relying on manually crafted agent roles and interaction prompts, wh…
arXiv:2604.16778v2 Announce Type: replace-cross Abstract: We propose a federated learning-like framework, Federation over Text (FoT), that enables multiple clients solving different tasks to collectively generate a shared library of metacognitive insights by iteratively federatin…
arXiv:2605.24426v1 Announce Type: new Abstract: Large Language Model (LLM) agents are increasingly improved through interaction, yet most self-evolution methods adapt either the policy or the learning environment in isolation. We identify this structural gap as \emph{Agent-Enviro…
arXiv:2605.25310v1 Announce Type: new Abstract: 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 struct…
arXiv cs.CL
TIER_1English(EN)·Daren Wang, Hong Xu, Jiawen Xian·
arXiv:2605.25958v1 Announce Type: new Abstract: This paper introduces PolyGnosis 2.0, a pioneering multi-agent architecture designed to extract predictive intelligence by synthesizing Polymarket anomaly signals with global Open Source Intelligence (OSINT) streams, specifically Gl…
arXiv cs.LG
TIER_1English(EN)·Ariel Fogel, Omer Hofman, Eilon Cohen, Roman Vainshtein·
arXiv:2602.04653v4 Announce Type: replace-cross Abstract: Open-weight language models are increasingly used in production settings, raising new security challenges. One prominent threat is backdoor attacks, in which adversaries embed hidden behaviors that activate under specific …
While large language models (LLMs) augmented with agentic search capabilities show promise for legal reasoning, they overlook a fundamental constraint that applicable law must match the temporal context of each case, as retroactive application of statutes violates core legal prin…
Agent-to-Agent (A2A) networks enable autonomous AI agents to collaborate by sharing reusable problem-solving instructions. However, how these decentralized ecosystems operate in practice remains largely unexplored. We present the first large-scale empirical study of EvoMap, a pro…
As large language model (LLM)-based multi-agent systems scale to handle increasingly complex tasks, balancing structural stability and dynamic adaptability becomes increasingly challenging. Existing systems typically adopt either structure-centric methods, committing to structure…
arXiv:2605.17076v2 Announce Type: replace-cross Abstract: We address concurrency control for LLM agents sharing mutable state over HTTP, where agents cannot be modified to declare read sets. S-Bus is an HTTP middleware whose central mechanism, a server-side DeliveryLog, reconstru…
arXiv:2605.23296v1 Announce Type: new Abstract: Long-horizon LLM agents accumulate growing conversation histories that eventually exceed the model's context window. Context compaction via LLM-based summarization keeps the conversation bounded, but summarization is inherently loss…
arXiv:2605.23887v1 Announce Type: cross Abstract: Temporal knowledge-graph data marketplaces face three coupled failures in static designs: stale hybrid index shortcuts reduce recall as edges evolve, stationary Shapley pricing misattributes value after distribution shifts, and un…
arXiv:2601.14652v5 Announce Type: replace Abstract: While multi-agent systems (MAS) promise elevated intelligence through coordination of agents, current approaches to automatic MAS design under-deliver. Such shortcomings stem from two key factors: (1) methodological complexity -…
arXiv:2605.18859v2 Announce Type: replace-cross Abstract: LLM routing matters most in long-horizon applications such as coding agents, deep research systems, and computer-use agents, where a single user request triggers many model calls. Routing each call to the cheapest sufficie…
arXiv:2605.23657v1 Announce Type: new Abstract: Skills, i.e., structured workflow instructions distilled for large language models (LLMs), are becoming an increasingly important mechanism for improving agent performance on real-world downstream tasks. However, as the open-source …
arXiv:2605.23574v1 Announce Type: new Abstract: Long-horizon language agents can make many plausible local tool calls yet fail to persist until a requested count is actually complete. We study this gap as Quantitative Goal Persistence (QGP): whether an agent keeps working until a…
Operating LLMs as coordinated multi-agent research systems over multi-hour runs surfaces failure modes that single-shot evaluation cannot: upstream providers throttle without warning, sub-agents drift the task to fit accessible tools, narrate machinery instead of using it, open r…
Large language model (LLM) agents excel at solving complex long-horizon tasks through autonomous interaction with environments. However, their real-world deployment faces a fundamental device--cloud dilemma: on-device models are efficient but often brittle, while cloud models are…
SEAL is a closed-loop co-evolution framework that simultaneously adapts both agent policies and training environments to improve interactive tool-use capabilities in large language models.
Temporal knowledge-graph data marketplaces face three coupled failures in static designs: stale hybrid index shortcuts reduce recall as edges evolve, stationary Shapley pricing misattributes value after distribution shifts, and uncoordinated agents over-consume a shared different…
Temporal knowledge-graph data marketplaces face three coupled failures in static designs: stale hybrid index shortcuts reduce recall as edges evolve, stationary Shapley pricing misattributes value after distribution shifts, and uncoordinated agents over-consume a shared different…
Skills, i.e., structured workflow instructions distilled for large language models (LLMs), are becoming an increasingly important mechanism for improving agent performance on real-world downstream tasks. However, as the open-source skill ecosystem rapidly expands, it remains uncl…
Long-horizon language agents can make many plausible local tool calls yet fail to persist until a requested count is actually complete. We study this gap as Quantitative Goal Persistence (QGP): whether an agent keeps working until an external verifier confirms enough distinct val…
Long-horizon LLM agents accumulate growing conversation histories that eventually exceed the model's context window. Context compaction via LLM-based summarization keeps the conversation bounded, but summarization is inherently lossy and the blocking call stalls agent inference f…
Topology optimization is a widely used design method that produces optimized material distributions for prescribed objectives and constraints through well-established numerical algorithms. Throughout the workflow, engineers make a series of decisions ranging from setting and adju…
arXiv:2605.20425v1 Announce Type: new Abstract: Designing multi-agent workflows is especially difficult in open-ended scientific settings where tasks lack curated training sets, reliable scalar evaluation metrics, and standardized interfaces between existing tools and agents. We …
arXiv:2605.22566v1 Announce Type: new Abstract: Large Language Model (LLM)-based agents demonstrate strong reasoning and execution capabilities on complex tasks when guided by structured instructions, commonly referred to as workflows. However, existing workflow-assisted agent se…
arXiv:2605.20923v1 Announce Type: cross Abstract: Distributed LLM agent workflows should not be monitored as if they produced a single sequential log. In an asynchronous execution, a decision can only depend on events that are causally visible to the lifeline that makes it: an ev…
In orchestrated multi-agent systems, humans often struggle to manage plans due to their complexity and limited transparency. Existing approaches rely on outcome-level supervision, where users verify only final outputs without visibility into intermediate reasoning. We formalize a…
Large Language Model (LLM)-based agents demonstrate strong reasoning and execution capabilities on complex tasks when guided by structured instructions, commonly referred to as workflows. However, existing workflow-assisted agent serving systems typically rely on predefined templ…
Most agent frameworks are built around the language model: a conversation loop comes first, then tools, then rules, and finally a logging layer bolted on for observability, with state persisted as retrievable "memory." We describe ActiveGraph, a runtime that inverts this arrangem…
Distributed LLM agent workflows should not be monitored as if they produced a single sequential log. In an asynchronous execution, a decision can only depend on events that are causally visible to the lifeline that makes it: an event that appears earlier in some log may still be …
Distributed LLM agent workflows should not be monitored as if they produced a single sequential log. In an asynchronous execution, a decision can only depend on events that are causally visible to the lifeline that makes it: an event that appears earlier in some log may still be …
arXiv cs.MA (Multiagent)
TIER_1English(EN)·Jason J. Choi·
Advanced Air Mobility (AAM) operations are expected to significantly increase aerial traffic in urban airspace, requiring autonomous traffic management systems to ensure collision-free operations in highly congested environments. In this paper, we propose a multi-agent coordinati…
Classical models of opinion dynamics assume human participants with bounded rationality and limited coordination. The rise of LLM-based agents introduces a qualitative shift: agents can now participate in online discussions at scale, maintain consistent persuasion strategies, and…
We introduce DecisionBench, a benchmark substrate for emergent delegation in long-horizon agentic workflows. The substrate fixes a task suite (GAIA, tau-bench, BFCL multi-turn), a peer-model pool (11 models, 7 vendor families), a delegation interface (call_model plus an optional …
The deployment of Large Language Models (LLMs) as autonomous economic agents introduces systemic risks that extend beyond individual capability failures. As agents transition to directly interacting with marketplaces, their collective behavior can amplify volatility and mask dece…
Multi-agent large language model (LLM) systems have shown promise for solving complex tasks through agent collaboration. However, existing frameworks assign tasks based on predefined roles without considering whether an agent can accurately assess its own competence boundaries, l…
We address concurrency control for LLM agents sharing mutable state over HTTP, where agents cannot be modified to declare read sets. S-Bus is an HTTP middleware whose central mechanism, a server-side DeliveryLog, reconstructs each agent's read set at commit time from observed HTT…
This is part 4 of a 6-part series we’re running about how product managers are using AI tools and vibe coding. Written by and for product managers. Summary Requirements docs, decks, and tickets go stale because PMs update them by hand. Agentic workflows fix the source of that pro…
Today, we’re excited to introduce Queue, a new capability designed to enhance the core Replit Agent experience. Queue allows users to submit multiple requests while the agent is actively working on a task, ensuring a continuous, uninterrupted app creation flow. As each task is co…
<p><em>How application observability extends to stochastic agent loops — and why the tool boundary matters.</em></p> <p>Production failures in LLM systems are often misattributed to the model. In practice, many incidents live in the <strong>action layer</strong>: a downstream API…
<p>Last month I spent three days debugging a Django service where the AI agent had written... mostly correct code. The endpoints worked. The tests passed. But somewhere around the fourth file, it had quietly dropped a database transaction wrapper around a multi-step write. By fil…
<!-- SC_OFF --><div class="md"><p>We've been trying to put LangGraph agents into production for a while. The thing that kept biting us was tool-call boundary enforcement: stuff like "must call X before Y", "max N retries", "approval gate before destructiv…
