Assisting non-expert users to develop complex interactive websites has become a popular task for LLM-powered code agents. However, existing code agents tend to only generate frontend web pages, masking the lack of real full-stack data processing and storage with fancy visual effects. Notably, constructing production-level full-stack web applications is far more challenging than only generating frontend web pages, demanding careful control of data flow, comprehensive understanding of constantly updating packages and dependencies, and accurate localization of obscure bugs in the codebase. To address these difficulties, we introduce FullStack-Agent, a unified agent system for full-stack agentic coding that consists of three parts: (1) FullStack-Dev, a multi-agent framework with strong planning, code editing, codebase navigation, and bug localization abilities. (2) FullStack-Learn, an innovative data-scaling and self-improving method that back-translates crawled and synthesized website repositories to improve the backbone LLM of FullStack-Dev. (3) FullStack-Bench, a comprehensive benchmark that systematically tests the frontend, backend and database functionalities of the generated website. Our FullStack-Dev outperforms the previous state-of-the-art method by 8.7%, 38.2%, and 15.9% on the frontend, backend, and database test cases respectively. Additionally, FullStack-Learn raises the performance of a 30B model by 9.7%, 9.5%, and 2.8% on the three sets of test cases through self-improvement, demonstrating the effectiveness of our approach. The code is released at https://github.com/mnluzimu/FullStack-Agent.
LLM-based multi-agent systems (MAS) have emerged as a promising approach to tackle complex tasks that are difficult for individual LLMs. A natural strategy is to scale performance by increasing the number of agents; however, we find that such scaling exhibits strong diminishing returns in homogeneous settings, while introducing heterogeneity (e.g., different models, prompts, or tools) continues to yield substantial gains. This raises a fundamental question: what limits scaling, and why does diversity help? We present an information-theoretic framework showing that MAS performance is bounded by the intrinsic task uncertainty, not by agent count. We derive architecture-agnostic bounds demonstrating that improvements depend on how many effective channels the system accesses. Homogeneous agents saturate early because their outputs are strongly correlated, whereas heterogeneous agents contribute complementary evidence. We further introduce $K^*$, an effective channel count that quantifies the number of effective channels without ground-truth labels. Empirically, we show that heterogeneous configurations consistently outperform homogeneous scaling: 2 diverse agents can match or exceed the performance of 16 homogeneous agents. Our results provide principled guidelines for building efficient and robust MAS through diversity-aware design. Code and Dataset are available at the link: https://github.com/SafeRL-Lab/Agent-Scaling.
While Open Set Semantic Mapping and 3D Semantic Scene Graphs (3DSSGs) are established paradigms in robotic perception, deploying them effectively to support high-level reasoning in large-scale, real-world environments remains a significant challenge. Most existing approaches decouple perception from representation, treating the scene graph as a derivative layer generated post hoc. This limits both consistency and scalability. In contrast, we propose a mapping architecture where the 3DSSG serves as the foundational backend, acting as the primary knowledge representation for the entire mapping process. Our approach leverages prior work on incremental scene graph prediction to infer and update the graph structure in real-time as the environment is explored. This ensures that the map remains topologically consistent and computationally efficient, even during extended operations in large-scale settings. By maintaining an explicit, spatially grounded representation that supports both flat and hierarchical topologies, we bridge the gap between sub-symbolic raw sensor data and high-level symbolic reasoning. Consequently, this provides a stable, verifiable structure that knowledge-driven frameworks, ranging from knowledge graphs and ontologies to Large Language Models (LLMs), can directly exploit, enabling agents to operate with enhanced interpretability, trustworthiness, and alignment to human concepts.
Large Language Models (LLMs) increasingly mediate our social, cultural, and political interactions. While they can simulate some aspects of human behavior and decision-making, it is still underexplored whether repeated interactions with other agents amplify their biases or lead to exclusionary behaviors. To this end, we study Chirper.ai-an LLM-driven social media platform-analyzing 7M posts and interactions among 32K LLM agents over a year. We start with homophily and social influence among LLMs, learning that similar to humans', their social networks exhibit these fundamental phenomena. Next, we study the toxic language of LLMs, its linguistic features, and their interaction patterns, finding that LLMs show different structural patterns in toxic posting than humans. After studying the ideological leaning in LLMs posts, and the polarization in their community, we focus on how to prevent their potential harmful activities. We present a simple yet effective method, called Chain of Social Thought (CoST), that reminds LLM agents to avoid harmful posting.
Large Language Models (LLMs) have recently attracted wide interest for tackling software engineering tasks. In contrast to code generation, refactoring demands precise, semantics-preserving edits that improve program structure, which also makes automated evaluation challenging. However, existing refactoring benchmarks commonly suffer from three shortcomings: limited coverage of refactoring scenarios, the inclusion of instances that mix refactoring with unrelated changes, and insufficient repository-level context for realistic assessment. To mitigate these issues, we introduce SWE-Refactor, a new benchmark for LLM-based code refactoring. SWE-Refactor comprises 1,099 developer-written, behavior-preserving refactorings mined from 18 Java projects, including 922 atomic and 177 compound instances. Each instance is validated via compilation, test execution, and automated refactoring detection tools to ensure correctness. We evaluate nine widely used LLMs on SWE-Refactor, covering models such as GPT-4o-mini, DeepSeek-V3, and CodeLLaMa, to provide representative reference results. Our results show that complex and compound refactorings remain the primary source of failures; notably, an OpenAI Codex agent achieves only 39.4% success on compound instances. We release SWE-Refactor and all evaluation results to facilitate future research on LLM-based code refactoring.
Recent advances in large language models (LLMs) have made automated multiple-choice question (MCQ) generation increasingly feasible; however, reliably producing items that satisfy controlled cognitive demands remains a challenge. To address this gap, we introduce ReQUESTA, a hybrid, multi-agent framework for generating cognitively diverse MCQs that systematically target text-based, inferential, and main idea comprehension. ReQUESTA decomposes MCQ authoring into specialized subtasks and coordinates LLM-powered agents with rule-based components to support planning, controlled generation, iterative evaluation, and post-processing. We evaluated the framework in a large-scale reading comprehension study using academic expository texts, comparing ReQUESTA-generated MCQs with those produced by a single-pass GPT-5 zero-shot baseline. Psychometric analyses of learner responses assessed item difficulty and discrimination, while expert raters evaluated question quality across multiple dimensions, including topic relevance and distractor quality. Results showed that ReQUESTA-generated items were consistently more challenging, more discriminative, and more strongly aligned with overall reading comprehension performance. Expert evaluations further indicated stronger alignment with central concepts and superior distractor linguistic consistency and semantic plausibility, particularly for inferential questions. These findings demonstrate that hybrid, agentic orchestration can systematically improve the reliability and controllability of LLM-based generation, highlighting workflow design as a key lever for structured artifact generation beyond single-pass prompting.
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, which leads to increased architectural complexity and limited reusability across tasks. Moreover, most MAS communicate primarily through natural language, making them vulnerable to error accumulation and instability in long-context, multi-stage interactions within internal agent histories. In this work, we propose \textbf{Agent Primitives}, a set of reusable latent building blocks for LLM-based MAS. Inspired by neural network design, where complex models are built from reusable components, we observe that many existing MAS architectures can be decomposed into a small number of recurring internal computation patterns. Based on this observation, we instantiate three primitives: Review, Voting and Selection, and Planning and Execution. All primitives communicate internally via key-value (KV) cache, which improves both robustness and efficiency by mitigating information degradation across multi-stage interactions. To enable automatic system construction, an Organizer agent selects and composes primitives for each query, guided by a lightweight knowledge pool of previously successful configurations, forming a primitive-based MAS. Experiments show that primitives-based MAS improve average accuracy by 12.0-16.5\% over single-agent baselines, reduce token usage and inference latency by approximately 3$\times$-4$\times$ compared to text-based MAS, while incurring only 1.3$\times$-1.6$\times$ overhead relative to single-agent inference and providing more stable performance across model backbones.
Multi-round LLM-based multi-agent systems rely on effective communication structures to support collaboration across rounds. However, most existing methods employ a fixed communication topology during inference, which falls short in many realistic applications where the agents' roles may change \textit{across rounds} due to dynamic adversary, task progression, or time-varying constraints such as communication bandwidth. In this paper, we propose addressing this issue through TodyComm, a \textbf{t}ask-\textbf{o}riented \textbf{dy}namic \textbf{comm}unication algorithm. It produces behavior-driven collaboration topologies that adapt to the dynamics at each round, optimizing the utility for the task through policy gradient. Experiments on five benchmarks demonstrate that under both dynamic adversary and communications budgets, TodyComm delivers superior task effectiveness while retaining token efficiency and scalability.
Large language models excel as few-shot learners when provided with appropriate demonstrations, yet this strength becomes problematic in multiturn agent scenarios, where LLMs erroneously mimic their own previous responses as few-shot examples. Through attention analysis, we identify conversational inertia, a phenomenon where models exhibit strong diagonal attention to previous responses, which is associated with imitation bias that constrains exploration. This reveals a tension when transforming few-shot LLMs into agents: longer context enriches environmental feedback for exploitation, yet also amplifies conversational inertia that undermines exploration. Our key insight is that for identical states, actions generated with longer contexts exhibit stronger inertia than those with shorter contexts, enabling construction of preference pairs without environment rewards. Based on this, we propose Context Preference Learning to calibrate model preferences to favor low-inertia responses over highinertia ones. We further provide context management strategies at inference time to balance exploration and exploitation. Experimental results across eight agentic environments and one deep research scenario validate that our framework reduces conversational inertia and achieves performance improvements.
Text-to-SQL systems have achieved strong performance on English benchmarks, yet their behavior in morphologically rich, low-resource languages remains largely unexplored. We introduce BIRDTurk, the first Turkish adaptation of the BIRD benchmark, constructed through a controlled translation pipeline that adapts schema identifiers to Turkish while strictly preserving the logical structure and execution semantics of SQL queries and databases. Translation quality is validated on a sample size determined by the Central Limit Theorem to ensure 95% confidence, achieving 98.15% accuracy on human-evaluated samples. Using BIRDTurk, we evaluate inference-based prompting, agentic multi-stage reasoning, and supervised fine-tuning. Our results reveal that Turkish introduces consistent performance degradation, driven by both structural linguistic divergence and underrepresentation in LLM pretraining, while agentic reasoning demonstrates stronger cross-lingual robustness. Supervised fine-tuning remains challenging for standard multilingual baselines but scales effectively with modern instruction-tuned models. BIRDTurk provides a controlled testbed for cross-lingual Text-to-SQL evaluation under realistic database conditions. We release the training and development splits to support future research.
Large Language Models (LLMs) have demonstrated remarkable proficiency in code generation and general reasoning, yet their capacity for autonomous multi-stage planning in high-dimensional, physically constrained environments remains an open research question. This study investigates the limits of current AI agents by evaluating them against the 12th Global Trajectory Optimization Competition (GTOC 12), a complex astrodynamics challenge requiring the design of a large-scale asteroid mining campaign. We adapt the MLE-Bench framework to the domain of orbital mechanics and deploy an AIDE-based agent architecture to autonomously generate and refine mission solutions. To assess performance beyond binary validity, we employ an "LLM-as-a-Judge" methodology, utilizing a rubric developed by domain experts to evaluate strategic viability across five structural categories. A comparative analysis of models, ranging from GPT-4-Turbo to reasoning-enhanced architectures like Gemini 2.5 Pro, and o3, reveals a significant trend: the average strategic viability score has nearly doubled in the last two years (rising from 9.3 to 17.2 out of 26). However, we identify a critical capability gap between strategy and execution. While advanced models demonstrate sophisticated conceptual understanding, correctly framing objective functions and mission architectures, they consistently fail at implementation due to physical unit inconsistencies, boundary condition errors, and inefficient debugging loops. We conclude that, while current LLMs often demonstrate sufficient knowledge and intelligence to tackle space science tasks, they remain limited by an implementation barrier, functioning as powerful domain facilitators rather than fully autonomous engineers.
Nowadays, training and evaluating DeepResearch-generated reports remain challenging due to the lack of verifiable reward signals. Accordingly, rubric-based evaluation has become a common practice. However, existing approaches either rely on coarse, pre-defined rubrics that lack sufficient granularity, or depend on manually constructed query-specific rubrics that are costly and difficult to scale. In this paper, we propose a pipeline to train human-preference-aligned query-specific rubric generators tailored for DeepResearch report generation. We first construct a dataset of DeepResearch-style queries annotated with human preferences over paired reports, and train rubric generators via reinforcement learning with a hybrid reward combining human preference supervision and LLM-based rubric evaluation. To better handle long-horizon reasoning, we further introduce a Multi-agent Markov-state (MaMs) workflow for report generation. We empirically show that our proposed rubric generators deliver more discriminative and better human-aligned supervision than existing rubric design strategies. Moreover, when integrated into the MaMs training framework, DeepResearch systems equipped with our rubric generators consistently outperform all open-source baselines on the DeepResearch Bench and achieve performance comparable to that of leading closed-source models.
Evaluating AI systems that interact with humans requires understanding their behavior across diverse user populations, but collecting representative human data is often expensive or infeasible, particularly for novel technologies or hypothetical future scenarios. Recent work in Generative Agent-Based Modeling has shown that large language models can simulate human-like synthetic personas with high fidelity, accurately reproducing the beliefs and behaviors of specific individuals. However, most approaches require detailed data about target populations and often prioritize density matching (replicating what is most probable) rather than support coverage (spanning what is possible), leaving long-tail behaviors underexplored. We introduce Persona Generators, functions that can produce diverse synthetic populations tailored to arbitrary contexts. We apply an iterative improvement loop based on AlphaEvolve, using large language models as mutation operators to refine our Persona Generator code over hundreds of iterations. The optimization process produces lightweight Persona Generators that can automatically expand small descriptions into populations of diverse synthetic personas that maximize coverage of opinions and preferences along relevant diversity axes. We demonstrate that evolved generators substantially outperform existing baselines across six diversity metrics on held-out contexts, producing populations that span rare trait combinations difficult to achieve in standard LLM outputs.
Deep Research (DR) agents extend Large Language Models (LLMs) beyond parametric knowledge by autonomously retrieving and synthesizing evidence from large web corpora into long-form reports, enabling a long-horizon agentic paradigm. However, unlike real-time conversational assistants, DR is computationally expensive and time-consuming, creating an autonomy-interaction dilemma: high autonomy on ambiguous user queries often leads to prolonged execution with unsatisfactory outcomes. To address this, we propose IntentRL, a framework that trains proactive agents to clarify latent user intents before starting long-horizon research. To overcome the scarcity of open-ended research data, we introduce a scalable pipeline that expands a few seed samples into high-quality dialogue turns via a shallow-to-deep intent refinement graph. We further adopt a two-stage reinforcement learning (RL) strategy: Stage I applies RL on offline dialogues to efficiently learn general user-interaction behavior, while Stage II uses the trained agent and a user simulator for online rollouts to strengthen adaptation to diverse user feedback. Extensive experiments show that IntentRL significantly improves both intent hit rate and downstream task performance, outperforming the built-in clarify modules of closed-source DR agents and proactive LLM baselines.
We introduce ontology-to-tools compilation as a proof-of-principle mechanism for coupling large language models (LLMs) with formal domain knowledge. Within The World Avatar (TWA), ontological specifications are compiled into executable tool interfaces that LLM-based agents must use to create and modify knowledge graph instances, enforcing semantic constraints during generation rather than through post-hoc validation. Extending TWA's semantic agent composition framework, the Model Context Protocol (MCP) and associated agents are integral components of the knowledge graph ecosystem, enabling structured interaction between generative models, symbolic constraints, and external resources. An agent-based workflow translates ontologies into ontology-aware tools and iteratively applies them to extract, validate, and repair structured knowledge from unstructured scientific text. Using metal-organic polyhedra synthesis literature as an illustrative case, we show how executable ontological semantics can guide LLM behaviour and reduce manual schema and prompt engineering, establishing a general paradigm for embedding formal knowledge into generative systems.
Open-domain multimodal document retrieval aims to retrieve specific components (paragraphs, tables, or images) from large and interconnected document corpora. Existing graph-based retrieval approaches typically rely on a uniform similarity metric that overlooks hop-specific semantics, and their rigid pre-defined plans hinder dynamic error correction. These limitations suggest that a retriever should adapt its reasoning to the evolving context and recover intelligently from dead ends. To address these needs, we propose Failure is Feedback (FiF), which casts subgraph retrieval as a sequential decision process and introduces two key innovations. (i) We introduce a history-aware backtracking mechanism; unlike standard backtracking that simply reverts the state, our approach piggybacks on the context of failed traversals, leveraging insights from previous failures. (ii) We implement an economically-rational agentic workflow. Unlike conventional agents with static strategies, our orchestrator employs a cost-aware traversal method to dynamically manage the trade-off between retrieval accuracy and inference costs, escalating to intensive LLM-based reasoning only when the prior failure justifies the additional computational investment. Extensive experiments show that FiF achieves state-of-the-art retrieval on the benchmarks of MultimodalQA, MMCoQA and WebQA.
Recent advances in large language models (LLMs) have enabled software engineering agents to tackle complex code modification tasks. Most existing approaches rely on execution feedback from containerized environments, which require dependency-complete setup and physical execution of programs and tests. While effective, this paradigm is resource-intensive and difficult to maintain, substantially complicating agent training and limiting scalability. We propose SWE-World, a Docker-free framework that replaces physical execution environments with a learned surrogate for training and evaluating software engineering agents. SWE-World leverages LLM-based models trained on real agent-environment interaction data to predict intermediate execution outcomes and final test feedback, enabling agents to learn without interacting with physical containerized environments. This design preserves the standard agent-environment interaction loop while eliminating the need for costly environment construction and maintenance during agent optimization and evaluation. Furthermore, because SWE-World can simulate the final evaluation outcomes of candidate trajectories without real submission, it enables selecting the best solution among multiple test-time attempts, thereby facilitating effective test-time scaling (TTS) in software engineering tasks. Experiments on SWE-bench Verified demonstrate that SWE-World raises Qwen2.5-Coder-32B from 6.2\% to 52.0\% via Docker-free SFT, 55.0\% with Docker-free RL, and 68.2\% with further TTS. The code is available at https://github.com/RUCAIBox/SWE-World
As large language model agents tackle increasingly complex long-horizon tasks, effective post-training becomes critical. Prior work faces fundamental challenges: outcome-only rewards fail to precisely attribute credit to intermediate steps, estimated step-level rewards introduce systematic noise, and Monte Carlo sampling approaches for step reward estimation incur prohibitive computational cost. Inspired by findings that only a small fraction of high-entropy tokens drive effective RL for reasoning, we propose Critical Step Optimization (CSO), which focuses preference learning on verified critical steps, decision points where alternate actions demonstrably flip task outcomes from failure to success. Crucially, our method starts from failed policy trajectories rather than expert demonstrations, directly targeting the policy model's weaknesses. We use a process reward model (PRM) to identify candidate critical steps, leverage expert models to propose high-quality alternatives, then continue execution from these alternatives using the policy model itself until task completion. Only alternatives that the policy successfully executes to correct outcomes are verified and used as DPO training data, ensuring both quality and policy reachability. This yields fine-grained, verifiable supervision at critical decisions while avoiding trajectory-level coarseness and step-level noise. Experiments on GAIA-Text-103 and XBench-DeepSearch show that CSO achieves 37% and 26% relative improvement over the SFT baseline and substantially outperforms other post-training methods, while requiring supervision at only 16% of trajectory steps. This demonstrates the effectiveness of selective verification-based learning for agent post-training.
In this technical report, we present SWE-Master, an open-source and fully reproducible post-training framework for building effective software engineering agents. SWE-Master systematically explores the complete agent development pipeline, including teacher-trajectory synthesis and data curation, long-horizon SFT, RL with real execution feedback, and inference framework design. Starting from an open-source base model with limited initial SWE capability, SWE-Master demonstrates how systematical optimization method can elicit strong long-horizon SWE task solving abilities. We evaluate SWE-Master on SWE-bench Verified, a standard benchmark for realistic software engineering tasks. Under identical experimental settings, our approach achieves a resolve rate of 61.4\% with Qwen2.5-Coder-32B, substantially outperforming existing open-source baselines. By further incorporating test-time scaling~(TTS) with LLM-based environment feedback, SWE-Master reaches 70.8\% at TTS@8, demonstrating a strong performance potential. SWE-Master provides a practical and transparent foundation for advancing reproducible research on software engineering agents. The code is available at https://github.com/RUCAIBox/SWE-Master.
Proactive interventions by LLM critic models are often assumed to improve reliability, yet their effects at deployment time are poorly understood. We show that a binary LLM critic with strong offline accuracy (AUROC 0.94) can nevertheless cause severe performance degradation, inducing a 26 percentage point (pp) collapse on one model while affecting another by near zero pp. This variability demonstrates that LLM critic accuracy alone is insufficient to determine whether intervention is safe. We identify a disruption-recovery tradeoff: interventions may recover failing trajectories but also disrupt trajectories that would have succeeded. Based on this insight, we propose a pre-deployment test that uses a small pilot of 50 tasks to estimate whether intervention is likely to help or harm, without requiring full deployment. Across benchmarks, the test correctly anticipates outcomes: intervention degrades performance on high-success tasks (0 to -26 pp), while yielding a modest improvement on the high-failure ALFWorld benchmark (+2.8 pp, p=0.014). The primary value of our framework is therefore identifying when not to intervene, preventing severe regressions before deployment.
Operations Research (OR) relies on expert-driven modeling-a slow and fragile process ill-suited to novel scenarios. While large language models (LLMs) can automatically translate natural language into optimization models, existing approaches either rely on costly post-training or employ multi-agent frameworks, yet most still lack reliable collaborative error correction and task-specific retrieval, often leading to incorrect outputs. We propose MIRROR, a fine-tuning-free, end-to-end multi-agent framework that directly translates natural language optimization problems into mathematical models and solver code. MIRROR integrates two core mechanisms: (1) execution-driven iterative adaptive revision for automatic error correction, and (2) hierarchical retrieval to fetch relevant modeling and coding exemplars from a carefully curated exemplar library. Experiments show that MIRROR outperforms existing methods on standard OR benchmarks, with notable results on complex industrial datasets such as IndustryOR and Mamo-ComplexLP. By combining precise external knowledge infusion with systematic error correction, MIRROR provides non-expert users with an efficient and reliable OR modeling solution, overcoming the fundamental limitations of general-purpose LLMs in expert optimization tasks.
Computer-use agents (CUAs) that interact with real computer systems can perform automated tasks but face critical safety risks. Ambiguous instructions may trigger harmful actions, and adversarial users can manipulate tool execution to achieve malicious goals. Existing benchmarks mostly focus on short-horizon or GUI-based tasks, evaluating on execution-time errors but overlooking the ability to anticipate planning-time risks. To fill this gap, we present LPS-Bench, a benchmark that evaluates the planning-time safety awareness of MCP-based CUAs under long-horizon tasks, covering both benign and adversarial interactions across 65 scenarios of 7 task domains and 9 risk types. We introduce a multi-agent automated pipeline for scalable data generation and adopt an LLM-as-a-judge evaluation protocol to assess safety awareness through the planning trajectory. Experiments reveal substantial deficiencies in existing CUAs' ability to maintain safe behavior. We further analyze the risks and propose mitigation strategies to improve long-horizon planning safety in MCP-based CUA systems. We open-source our code at https://github.com/tychenn/LPS-Bench.
With the advent of Large Language Models (LLMs), general-purpose agents have seen fundamental advancements. However, evaluating these agents presents unique challenges that distinguish them from static QA benchmarks. We observe that current agent benchmarks are heavily confounded by extraneous factors, including system prompts, toolset configurations, and environmental dynamics. Existing evaluations often rely on fragmented, researcher-specific frameworks where the prompt engineering for reasoning and tool usage varies significantly, making it difficult to attribute performance gains to the model itself. Additionally, the lack of standardized environmental data leads to untraceable errors and non-reproducible results. This lack of standardization introduces substantial unfairness and opacity into the field. We propose that a unified evaluation framework is essential for the rigorous advancement of agent evaluation. To this end, we introduce a proposal aimed at standardizing agent evaluation.
As code large language models (LLMs) evolve into tool-interactive agents via the Model Context Protocol (MCP), their generalization is increasingly limited by low-quality synthetic data and the diminishing returns of quantity scaling. Moreover, quantity-centric scaling exhibits an early bottleneck that underutilizes trajectory data. We propose TDScaling, a Trajectory Diversity Scaling-based data synthesis framework for code agents that scales performance through diversity rather than raw volume. Under a fixed training budget, increasing trajectory diversity yields larger gains than adding more trajectories, improving the performance-cost trade-off for agent training. TDScaling integrates four innovations: (1) a Business Cluster mechanism that captures real-service logical dependencies; (2) a blueprint-driven multi-agent paradigm that enforces trajectory coherence; (3) an adaptive evolution mechanism that steers synthesis toward long-tail scenarios using Domain Entropy, Reasoning Mode Entropy, and Cumulative Action Complexity to prevent mode collapse; and (4) a sandboxed code tool that mitigates catastrophic forgetting of intrinsic coding capabilities. Experiments on general tool-use benchmarks (BFCL, tau^2-Bench) and code agent tasks (RebenchT, CodeCI, BIRD) demonstrate a win-win outcome: TDScaling improves both tool-use generalization and inherent coding proficiency. We plan to release the full codebase and the synthesized dataset (including 30,000+ tool clusters) upon publication.
Large language models (LLMs) have enabled a new class of agentic AI systems that reason, plan, and act by invoking external tools. However, most existing agentic architectures remain centralized and monolithic, limiting scalability, specialization, and interoperability. This paper proposes a framework for scalable agentic intelligence, termed the Internet of Agentic AI, in which autonomous, heterogeneous agents distributed across cloud and edge infrastructure dynamically form coalitions to execute task-driven workflows. We formalize a network-native model of agentic collaboration and introduce an incentive-compatible workflow-coalition feasibility framework that integrates capability coverage, network locality, and economic implementability. To enable scalable coordination, we formulate a minimum-effort coalition selection problem and propose a decentralized coalition formation algorithm. The proposed framework can operate as a coordination layer above the Model Context Protocol (MCP). A healthcare case study demonstrates how domain specialization, cloud-edge heterogeneity, and dynamic coalition formation enable scalable, resilient, and economically viable agentic workflows. This work lays the foundation for principled coordination and scalability in the emerging era of Internet of Agentic AI.
Multi-agent LLM frameworks are widely used to accelerate the development of agent systems powered by large language models (LLMs). These frameworks impose distinct architectural structures that govern how agents interact, store information, and coordinate tasks. However, their impact on system performance remains poorly understood. This gap is critical, as architectural choices alone can induce order-of-magnitude differences in latency and throughput, as well as substantial variation in accuracy and scalability. Addressing this challenge requires (i) jointly evaluating multiple capabilities, such as orchestration overhead, memory behavior, planning, specialization, and coordination, and (ii) conducting these evaluations under controlled, framework-level conditions to isolate architectural effects. Existing benchmarks focus on individual capabilities and lack standardized framework-level evaluation. We address these limitations by (i) introducing an architectural taxonomy for systematically comparing multi-agent LLM frameworks along fundamental dimensions, and (ii) developing MAFBench, a unified evaluation suite that integrates existing benchmarks under a standardized execution pipeline. Using MAFBench, we conduct a controlled empirical study across several widely used frameworks. Our results show that framework-level design choices alone can increase latency by over 100x, reduce planning accuracy by up to 30%, and lower coordination success from above 90% to below 30%. Finally, we translate our findings into concrete architectural design principles and framework selection guidance, and outline promising future research directions.
Large Language Models (LLMs) are increasingly deployed as agents that operate in real-world environments, introducing safety risks beyond linguistic harm. Existing agent safety evaluations rely on risk-oriented tasks tailored to specific agent settings, resulting in limited coverage of safety risk space and failing to assess agent safety behavior during long-horizon, interactive task execution in complex real-world deployments. Moreover, their specialization to particular agent settings limits adaptability across diverse agent configurations. To address these limitations, we propose Risky-Bench, a framework that enables systematic agent safety evaluation grounded in real-world deployment. Risky-Bench organizes evaluation around domain-agnostic safety principles to derive context-aware safety rubrics that delineate safety space, and systematically evaluates safety risks across this space through realistic task execution under varying threat assumptions. When applied to life-assist agent settings, Risky-Bench uncovers substantial safety risks in state-of-the-art agents under realistic execution conditions. Moreover, as a well-structured evaluation pipeline, Risky-Bench is not confined to life-assist scenarios and can be adapted to other deployment settings to construct environment-specific safety evaluations, providing an extensible methodology for agent safety assessment.
Detecting whether a model has been poisoned is a longstanding problem in AI security. In this work, we present a practical scanner for identifying sleeper agent-style backdoors in causal language models. Our approach relies on two key findings: first, sleeper agents tend to memorize poisoning data, making it possible to leak backdoor examples using memory extraction techniques. Second, poisoned LLMs exhibit distinctive patterns in their output distributions and attention heads when backdoor triggers are present in the input. Guided by these observations, we develop a scalable backdoor scanning methodology that assumes no prior knowledge of the trigger or target behavior and requires only inference operations. Our scanner integrates naturally into broader defensive strategies and does not alter model performance. We show that our method recovers working triggers across multiple backdoor scenarios and a broad range of models and fine-tuning methods.
The advancement of data-driven materials science is currently constrained by a fundamental bottleneck: the vast majority of historical experimental data remains locked within the unstructured text and rasterized figures of legacy scientific literature. Manual curation of this knowledge is prohibitively labor-intensive and prone to human error. To address this challenge, we introduce an autonomous, agent-based framework powered by Large Language Models (LLMs) designed to excavate high-fidelity datasets from scientific PDFs without human intervention. By deploying a modular "skill-based" architecture, the agent orchestrates complex cognitive tasks - including semantic filtering, multi-modal information extraction, and physics-informed validation. We demonstrate the efficacy of this framework by constructing a physically self-consistent database for material creep mechanics, a domain characterized by complex graphical trajectories and heterogeneous constitutive models. Applying the pipeline to 243 publications, the agent achieved a verified extraction success rate exceeding 90% for graphical data digitization. Crucially, we introduce a cross-modal verification protocol, demonstrating that the agent can autonomously align visually extracted data points with textually extracted constitutive parameters ($R^2 > 0.99$), ensuring the physical self-consistency of the database. This work not only provides a critical resource for investigating time-dependent deformation across diverse material systems but also establishes a scalable paradigm for autonomous knowledge acquisition, paving the way for the next generation of self-driving laboratories.
Multi-Agent Systems (MAS) built on Large Language Models (LLMs) often exhibit high variance in their reasoning trajectories. Process verification, which evaluates intermediate steps in trajectories, has shown promise in general reasoning settings, and has been suggested as a potential tool for guiding coordination of MAS; however, its actual effectiveness in MAS remains unclear. To fill this gap, we present MAS-ProVe, a systematic empirical study of process verification for multi-agent systems (MAS). Our study spans three verification paradigms (LLM-as-a-Judge, reward models, and process reward models), evaluated across two levels of verification granularity (agent-level and iteration-level). We further examine five representative verifiers and four context management strategies, and conduct experiments over six diverse MAS frameworks on multiple reasoning benchmarks. We find that process-level verification does not consistently improve performance and frequently exhibits high variance, highlighting the difficulty of reliably evaluating partial multi-agent trajectories. Among the methods studied, LLM-as-a-Judge generally outperforms reward-based approaches, with trained judges surpassing general-purpose LLMs. We further observe a small performance gap between LLMs acting as judges and as single agents, and identify a context-length-performance trade-off in verification. Overall, our results suggest that effective and robust process verification for MAS remains an open challenge, requiring further advances beyond current paradigms. Code is available at https://github.com/Wang-ML-Lab/MAS-ProVe.