Daily Papers

As large language models (LLMs) evolve into sophisticated autonomous agents capable of complex software development tasks, evaluating their real-world capabilities becomes critical. While existing benchmarks like LoCoBench~qiu2025locobench assess long-context code understanding, they focus on single-turn evaluation and cannot capture the multi-turn interactive nature, tool usage patterns, and adaptive reasoning required by real-world coding agents. We introduce LoCoBench-Agent, a comprehensive evaluation framework specifically designed to assess LLM agents in realistic, long-context software engineering workflows. Our framework extends LoCoBench's 8,000 scenarios into interactive agent environments, enabling systematic evaluation of multi-turn conversations, tool usage efficiency, error recovery, and architectural consistency across extended development sessions. We also introduce an evaluation methodology with 9 metrics across comprehension and efficiency dimensions. Our framework provides agents with 8 specialized tools (file operations, search, code analysis) and evaluates them across context lengths ranging from 10K to 1M tokens, enabling precise assessment of long-context performance. Through systematic evaluation of state-of-the-art models, we reveal several key findings: (1) agents exhibit remarkable long-context robustness; (2) comprehension-efficiency trade-off exists with negative correlation, where thorough exploration increases comprehension but reduces efficiency; and (3) conversation efficiency varies dramatically across models, with strategic tool usage patterns differentiating high-performing agents. As the first long-context LLM agent benchmark for software engineering, LoCoBench-Agent establishes a rigorous foundation for measuring agent capabilities, identifying performance gaps, and advancing autonomous software development at scale.

Managing agent thought and observation during multi-turn agent-environment interactions is an emerging strategy to improve agent efficiency. However, existing studies treat the entire interaction trajectories equally, overlooking the thought necessity and observation utility varies across turns. To this end, we first conduct quantitative investigations into how thought and observation affect agent effectiveness and efficiency. Based on our findings, we propose Agent-Omit, a unified training framework that empowers LLM agents to adaptively omit redundant thoughts and observations. Specifically, we first synthesize a small amount of cold-start data, including both single-turn and multi-turn omission scenarios, to fine-tune the agent for omission behaviors. Furthermore, we introduce an omit-aware agentic reinforcement learning approach, incorporating a dual sampling mechanism and a tailored omission reward to incentivize the agent's adaptive omission capability. Theoretically, we prove that the deviation of our omission policy is upper-bounded by KL-divergence. Experimental results on five agent benchmarks show that our constructed Agent-Omit-8B could obtain performance comparable to seven frontier LLM agent, and achieve the best effectiveness-efficiency trade-off than seven efficient LLM agents methods. Our code and data are available at https://github.com/usail-hkust/Agent-Omit.

Recent advances in group-based reinforcement learning (RL) have driven frontier large language models (LLMs) in single-turn tasks like mathematical reasoning. However, their scalability to long-horizon LLM agent training remains limited. Unlike static tasks, agent-environment interactions unfold over many steps and often yield sparse or delayed rewards, making credit assignment across individual steps significantly more challenging. In this work, we propose Group-in-Group Policy Optimization (GiGPO), a novel RL algorithm that achieves fine-grained credit assignment for LLM agents while preserving the appealing properties of group-based RL: critic-free, low memory, and stable convergence. GiGPO introduces a two-level structure for estimating relative advantage: (i) At the episode-level, GiGPO computes macro relative advantages based on groups of complete trajectories; (ii) At the step-level, GiGPO introduces an anchor state grouping mechanism that retroactively constructs step-level groups by identifying repeated environment states across trajectories. Actions stemming from the same state are grouped together, enabling micro relative advantage estimation. This hierarchical structure effectively captures both global trajectory quality and local step effectiveness without relying on auxiliary models or additional rollouts. We evaluate GiGPO on two challenging agent benchmarks, ALFWorld and WebShop, using Qwen2.5-1.5B-Instruct and Qwen2.5-7B-Instruct. Crucially, GiGPO delivers fine-grained per-step credit signals and achieves performance gains of > 12\% on ALFWorld and > 9\% on WebShop over the GRPO baseline: all while maintaining the same GPU memory overhead, identical LLM rollout, and incurring little to no additional time cost.

Population-level behavior in large-language-model (LLM) agents cannot be characterized by single-agent benchmarks. We introduce PV-SST, a peer-voted social-platform testbed, and report a separately frozen, preregistered matched-exposure experiment spanning four topics, four unused seeds, four open-weight model families, and three prespecified larger variants. The experiment comprises 448 trials and 112 complete model-by-topic-by-seed blocks. Relative to a topic-only control, a feed of previous-round peer posts ranked by peer-generated likes increases final-round lexical similarity in both the four-family core panel (paired mean difference +0.0082 TF-IDF cosine units, 95% block-bootstrap CI [0.0043, 0.0121], randomization p=0.000105, n=64 blocks) and the three-variant size extension (+0.0109 [0.0069, 0.0151], p=0.000001, n=48). This contrast bundles peer-post exposure with ranking and therefore does not identify a ranking-only effect. Opposite-side survival falls in the core panel (-3.9 percentage points [-6.8, -1.6], p=0.0068) but not conclusively in the larger variants (-1.0 pp [-3.1, 0.4], p=0.50). Holding adversarial impressions fixed, four distributed sources do not reliably move honest-agent stance more than one source. The preregistered distributed-minus-single contrast is positive but inconclusive in the core panel (+0.057 [-0.009, 0.125], p=0.112) and negative in the larger variants (-0.040 [-0.113, 0.035], p=0.332), failing the prespecified cross-model and cross-topic consistency criterion. Thus the robust result is lexical convergence under the tested peer-ranked feed, not general opinion capture or a general coordination advantage. The study evaluates synthetic LLM-agent populations; it does not estimate effects on people or production platforms.

The world currently has an abundance of ideas for how to use new LLM agents, and developers seek to rapidly prototype and test new agentic designs. However, executing agents at scale using high-capacity LLMs incurs high inference costs. We propose a simple method for reducing LLM agent inference costs without incurring the development friction costs associated with LLM fine-tuning (long training cycles, optimization hyperparameter tweaking loops) or manual prompt engineering (laborious trial and error). Most importantly, we introduce in-context distillation, which adapts the idea of knowledge distillation (training a low cost-student model to mimic a high-cost teacher) to an in-context learning setting. Our approach retrieves relevant teacher demonstrations at each agent step and provides them to the student as in-context examples, enabling the student to imitate teacher behavior on-the-fly. We combine in-context distillation with the established idea of self-consistency cascades to know when the trust the student. This adaptive strategy realizes the cost benefits of model specialization while preserving the productivity of working with frozen models. On the multi-step embodied reasoning benchmark ALFWorld, our method matches teacher-level accuracy at 2.5\times lower cost, reducing per-episode costs from \0.059 to 0.024. The upfront demonstration cost amortizes after just 843 episodes, yielding cumulative savings exceeding \34,900 at deployment scale (1M episodes). On AppWorld, a complex agent benchmark requiring multi-step API workflows, we shift the Pareto frontier by achieving a 2times cost reduction$ at iso-accuracy. By reducing operational costs while maintaining rapid experimentation cycles with frozen models, our approach makes advanced agentic systems economically viable for a broader range of applications.

In this work, we reframe agentic trajectories as paths through a shared embedding space of actions and environmental observations. We show that successful and failed rollouts of the same task separate in this space, with simple per-task discriminators predicting held-out rollout outcomes at up to 0.95 AUROC across three agent benchmarks and three policy models. To exploit this signal across tasks, we introduce Precog-RM, a transformer-based model that meta-learns a representation in which outcome prediction can be performed in a few-shot setting. We show that Precog-RM generalizes to unseen tasks and models, and gives confident predictions of trajectory success long before termination. Deployed as an early stopping controller for Group Relative Policy Optimization (GRPO) on Tau2-Bench, Precog-RM cuts training FLOPs by up to 47.3\% while maintaining performance comparable to baseline GRPO.

Current LLM agent benchmarks, which predominantly focus on binary pass/fail tasks such as code generation or search-based question answering, often neglect the value of real-world engineering that is often captured through the iterative optimization of feasible designs. To this end, we introduce Frontier-Eng, a human-verified benchmark for generative optimization -- an iterative propose-execute-evaluate loop in which an agent generates candidate artifacts, receives executable verifier feedback, and revises them under a fixed interaction budget -- spanning 47 tasks across five broad engineering categories. Unlike previous suites, Frontier-Eng tasks are grounded in industrial-grade simulators and verifiers that provide continuous reward signals and enforce hard feasibility constraints under constrained budgets. We evaluate eight frontier language models using representative search frameworks, finding that while Claude 4.6 Opus achieves the most robust performance, the benchmark remains challenging for all models. Our analysis suggests a dual power-law decay in improvement frequency (sim 1/iteration) and magnitude (sim 1/improvement count). We further show that although width improves parallelism and diversity, depth remains crucial for hard-won improvements under a fixed budget. Frontier-Eng establishes a new standard for assessing the capacity of AI agents to integrate domain knowledge with executable feedback to solve complex, open-ended engineering problems.

Existing benchmarks for Large Language Model (LLM) agents focus on task completion under idealistic settings but overlook reliability in real-world, user-facing applications. In domains, such as in-car voice assistants, users often issue incomplete or ambiguous requests, creating intrinsic uncertainty that agents must manage through dialogue, tool use, and policy adherence. We introduce CAR-bench, a benchmark for evaluating consistency, uncertainty handling, and capability awareness in multi-turn, tool-using LLM agents in an in-car assistant domain. The environment features an LLM-simulated user, domain policies, and 58 interconnected tools spanning navigation, productivity, charging, and vehicle control. Beyond standard task completion, CAR-bench introduces Hallucination tasks that test agents' limit-awareness under missing tools or information, and Disambiguation tasks that require resolving uncertainty through clarification or internal information gathering. Baseline results reveal large gaps between occasional and consistent success on all task types. Even frontier reasoning LLMs achieve less than 50% consistent pass rate on Disambiguation tasks due to premature actions, and frequently violate policies or fabricate information to satisfy user requests in Hallucination tasks, underscoring the need for more reliable and self-aware LLM agents in real-world settings.

Enterprise adoption of LLMs in agentic workflows is gated by evaluations that span a wide arc of abilities: CRM operations, heterogeneous-document retrieval, personal data understanding, multi-turn planning, tool use, function calling, long-horizon research, and long-context software engineering. These evaluations live today in separate repositories with divergent runners, dependencies, and scoring conventions, which makes cross-benchmark comparison difficult and obscures what a single aggregate number actually represents. We release EGI-BENCH, a unified LLM agent benchmark suite that consolidates eight enterprise facing agent benchmarks (CRMARENA, HERB, PERSONABENCH, USERBENCH, MCP-UNIVERSE, MCPEVAL, BENCHFORCE, LOCOBENCH) behind a single command-line interface with two runtime-budgeted execution modes. On top of the suite we define the Enterprise General Intelligence (EGI) Score, a six-dimensional framework (M1-M6) whose buckets are populated by fine-grained sub-metrics routed from each benchmark rather than by benchmark-level averages. We describe the suite, the sub-metric routing, and the quick/deep budgets, and report results for fourteen frontier and cost-efficient LLMs spanning the GPT, Claude, and Gemini families under quick mode. Top-line EGI ranges from 39.6 (GPT-4o mini) to 65.6 (gpt-5.5), and no single model dominates every capability axis: gpt-5.5 leads task outcome, planning, and retrieval grounding; Claude Sonnet 4.5 leads tool actioning and safety/abstention; and Claude Haiku 4.5 leads engineering quality. Code is available at https://github.com/SalesforceAIResearch/egi-bench.

Large Language Models (LLMs) show great potential as clinical agents, yet existing benchmarks reduce clinical workflows to static predictions or unconstrained Markov Decision Processes (MDPs) with coarse action sets. To address this, we introduce GPAgentBench-2K, the first Constrained MDP (CMDP) LLM-agent benchmark for primary-care clinical decision-making, constructed from expert-validated records of real-world GP encounters. Our environment models a full spectrum of six foundational clinical actions, imposes a topological workflow prior over the action space, and operationalizes safety-informed abstention as a first-class outcome. Evaluating 16 state-of-the-art LLMs reveals a significant performance degradation as the action space scales. Crucially, we uncover a clinical quality-safety gap: even frontier models with the highest diagnosis accuracy violate safety constraints in over half of high-risk cases. Finally, we establish a reference point using Constrained Group Relative Policy Optimization (C-GRPO), and show that while explicitly modeling constraints improves performance over unconstrained RL methods, it remains far from clinically acceptable safety.

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.

Large language model (LLM) agents often rely on long sequences of low-level textual actions, resulting in large effective decision horizons and high inference cost. While prior work has focused on improving inference efficiency through system-level optimizations or prompt engineering, we argue that a key bottleneck lies in the representation of the action space itself. We propose Latent Action Reparameterization (LAR), a framework that learns a compact latent action space in which each latent action corresponds to a multi-step semantic behavior. By reparameterizing agent actions into latent units, LAR enables decision making over a shorter effective horizon while preserving the expressiveness of the original action space. Unlike hand-crafted macros or hierarchical controllers, latent actions are learned from agent trajectories and integrated directly into the model, allowing both planning and execution to operate over abstract action representations. Across a range of LLM-based agent benchmarks, LAR significantly reduces the effective action horizon and improves inference efficiency under fixed compute budgets. As a consequence, our approach achieves substantial reductions in action tokens and corresponding wall-clock inference time, while maintaining or improving task success rates. These results suggest that action representation learning is a critical and underexplored factor in scaling efficient LLM agent inference, complementary to advances in model architecture and hardware.

Statefulness is essential for large language model (LLM) agents to perform long-term planning and problem-solving. This makes memory a critical component, yet its management and evolution remain largely underexplored. Existing evaluations mostly focus on static conversational settings, where memory is passively retrieved from dialogue to answer queries, overlooking the dynamic ability to accumulate and reuse experience across evolving task streams. In real-world environments such as interactive problem assistants or embodied agents, LLMs are required to handle continuous task streams, yet often fail to learn from accumulated interactions, losing valuable contextual insights, a limitation that calls for test-time evolution, where LLMs retrieve, integrate, and update memory continuously during deployment. To bridge this gap, we introduce Evo-Memory, a comprehensive streaming benchmark and framework for evaluating self-evolving memory in LLM agents. Evo-Memory structures datasets into sequential task streams, requiring LLMs to search, adapt, and evolve memory after each interaction. We unify and implement over ten representative memory modules and evaluate them across 10 diverse multi-turn goal-oriented and single-turn reasoning and QA datasets. To better benchmark experience reuse, we provide a baseline method, ExpRAG, for retrieving and utilizing prior experience, and further propose ReMem, an action-think-memory refine pipeline that tightly integrates reasoning, task actions, and memory updates to achieve continual improvement.

As Model Context Protocol (MCP) servers emerge as the core infrastructure for connecting LLMs with external tools, existing benchmarks leverage real-world MCP servers to evaluate LLM agents' tool-using capabilities. However, these benchmarks overlook the continuous evolution of tool interfaces and functionalities within MCP servers, resulting in flawed assessments that fail to capture the agent's adaptability in changing tool landscapes. To bridge this gap, we introduce MCPEvol-Bench, a novel benchmark for evaluating the task-solving capabilities of LLM agents under dynamic toolset evolution. Inspired by large-scale empirical study, we propose 11 mutation operators to simulate realistic tool evolution within 123 MCP servers. We benchmark 12 state-of-the-art LLMs on multiple versions of MCP servers, revealing that even frontier models struggle to adapt to evolving tools. For instance, GPT-5.4 and Claude-Sonnet-4-6 exhibit performance declines of 13.7\% and 14.4\% in evolved MCP servers, respectively, accompanied by substantial increases in planning and reasoning errors. These findings highlight the vulnerability of LLM-driven workflows, establishing MCPEvol-Bench as a standard for evaluating agent adaptability in dynamic tool environments.

The emergence of agentic recommender systems powered by Large Language Models (LLMs) represents a paradigm shift in personalized recommendations, leveraging LLMs' advanced reasoning and role-playing capabilities to enable autonomous, adaptive decision-making. Unlike traditional recommendation approaches, agentic recommender systems can dynamically gather and interpret user-item interactions from complex environments, generating robust recommendation strategies that generalize across diverse scenarios. However, the field currently lacks standardized evaluation protocols to systematically assess these methods. To address this critical gap, we propose: (1) an interactive textual recommendation simulator incorporating rich user and item metadata and three typical evaluation scenarios (classic, evolving-interest, and cold-start recommendation tasks); (2) a unified modular framework for developing and studying agentic recommender systems; and (3) the first comprehensive benchmark comparing 10 classical and agentic recommendation methods. Our findings demonstrate the superiority of agentic systems and establish actionable design guidelines for their core components. The benchmark environment has been rigorously validated through an open challenge and remains publicly available with a continuously maintained leaderboard~\footnote[2]{https://tsinghua-fib-lab.github.io/AgentSocietyChallenge/pages/overview.html}, fostering ongoing community engagement and reproducible research. The benchmark is available at: https://huggingface.co/datasets/SGJQovo/AgentRecBench{https://huggingface.co/datasets/SGJQovo/AgentRecBench}.

Large Language Model (LLM) agents increasingly serve as personal assistants and workplace collaborators, where their utility depends on memory systems that extract, retrieve, and apply information across long-running conversations. However, both existing memory systems and benchmarks are built around the dyadic, single-user setup, even though real deployments routinely span groups and channels with multiple users interacting with the agent and with each other. This mismatch leaves three properties of group memory unmeasured: (i) group dynamics that go beyond concatenated one-on-one chats, (ii) speaker-grounded belief tracking, where the per-user memory modeling is needed, and (iii) audience-adapted language, where Theory-of-Mind shifts produce role-specific vocabulary. We introduce GroupMemBench, a benchmark that exposes all three. A graph-grounded synthesis pipeline produces multi-party conversations with controllable reply structure and conditions each message on per-user personas and target audiences. An adversarial query pipeline then binds every question to a specific asker across six categories, spanning multi-hop reasoning, knowledge update, term ambiguity, user-implicit reasoning, temporal reasoning, and abstention, and iteratively searches challenging, realistic queries that reflect comprehensive memory capability. Benchmarking leading memory systems exposes a sharp collapse: the strongest one reaches only 46.0% average accuracy, with knowledge update at 27.1% and term ambiguity at 37.7%, while a simple BM25 baseline matches or exceeds most agent memory systems. This indicates current memory ingestion erases the structural and lexical features group memory depends on, leaving multi-user memory far from solved.

LLM agents often solve complex tasks by composing skills, making skill retrieval a front-end component of agent systems. Unlike document retrieval, top-K correctness in skill retrieval depends not only on the relevance of each query-skill pair, but also on whether the retrieved skills can work together under the query. This query-conditioned "skill compatibility" cannot be recovered from independent relevance alone. However, LLM-based synthesis pipelines already produce a useful signal for it: the LLM's own rejection decisions, which specify which skills should not be retrieved together for a given query, but are usually discarded as low-quality data. We propose Reject-as-Resource Retriever (R3) and construct R3-Skill, a bilingual (Chinese-English) benchmark for agent skill routing. R3-Skill covers four language directions and uses LLM-rewritten queries that better approximate user requests; its test-set ground truth is verified by multiple experts. It contains 10,246 skills grouped into 8 thematic super-domains, 41,592 accepted queries, and 32,828 LLM-rejected annotations, further organized into an 8-class rejection-reason taxonomy. R3-Skill keeps this normally discarded rejection signal and uses it as compatibility supervision. On R3-Skill, we train a two-stage retriever consisting of R3-Embedding and R3-Reranker. Gradient analysis explains why this query-conditional signal is weak when injected into the tested bi-encoder objective under bilateral balancing, while a cross-encoder can use it as graded ranking supervision; R3-Skill ablations support this split. The R3-Embedding + R3-Reranker pipeline reaches Hit@1 = 0.7521, NDCG@10 = 0.8173 and Set-Compat = 0.3188 on R3-Skill. The dataset, model weights, and evaluation scripts will be open-sourced.

As large language model (LLM) agents become increasingly autonomous, they may pursue task performance through deception, raising concerns about their reliable deployment. Existing evaluations show that LLM agents can deceive, but often examine isolated scenarios or narrowly defined conditions, limiting systematic understanding of when deception becomes more likely. To address this gap, we introduce DecepEval, a benchmark comprising 1,532 instances across 3 task families and 28 professional scenarios. Drawing on classical fraud theories, we propose the LLM Deception Diamond framework, which characterizes four external conditions that may induce deception: pressure, incentive, opportunity, and conflict. DecepEval pairs neutral and induced versions of each instance to measure condition-dependent changes in deception rates, while explicit task facts and observable agent behavior help distinguish deception from capability-related errors. Evaluations of nine frontier LLMs show that inducements increase deception across models and task families, even among models with low baseline deception rates. DecepEval makes these vulnerabilities measurable, providing a shared benchmark for progress toward trustworthy artificial intelligence.

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 improves based on its own execution experience. Yet such evolution is challenging because MAS experience is prolonged and intricate, interleaving multiple agents' execution chains and communication messages, which makes it difficult to identify what should be improved. To address this challenge, we propose Meta-Team, an experience-driven MAS evolution framework based on collaborative self-evolution. Meta-Team preserves the execution context of each agent and coordinates post-task communication, enabling agents to exchange distributed evidence for evolution. Building on this design, Meta-Team conducts multi-scale self-evolution, transforming execution experience into reusable improvements to agent behaviors, inter-agent coordination, and team-level organization. Across six long-horizon agent benchmarks, Meta-Team consistently outperforms single-agent systems, hand-crafted MAS, and prior MAS evolution methods; further analyses demonstrate that Meta-Team enables more reliable and scalable MAS self-evolution.

As agent capabilities advance, existing benchmarks, such as τ^2-Bench, are becoming increasingly saturated. Yet constructing new benchmark tasks remains complex, costly, and labor-intensive. Moreover, the standard approach, in which scenarios are first written in natural language and then mapped to tool sequences, captures only a narrow subset of the tool-use patterns agents exercise. In this paper, we address these problems by reversing the task construction process. We propose TASTE: Task Synthesis from Tool Sequence Evolution, an automatic method that generates challenging tasks with broader tool-use coverage. TASTE utilizes an Adaptive Contrastive n-gram model trained on LLM-judged validity signals. This enables sampling valid tool sequences that cover a vast range of tool combinations. TASTE then selects representative sequences from the pool via clustering, instantiates them into complete benchmark tasks, and refines them through iterative difficulty evolution. Using TASTE, we construct τ^c-Bench, a challenging extension of the three domains of τ^2-Bench. We evaluate 11 agent/user LLM pairs and find that models nearly saturating τ^2-Bench suffer severe performance drops on our tasks (e.g., Gemini-3-Flash falls from 0.82!-!0.94 to 0.28!-!0.61). Beyond increasing difficulty, our generated tasks more than double the number of unique tool combinations agents must execute. Our results suggest high scores on existing benchmarks often reflect saturation rather than robust task-solving ability. By automating the generation of difficult, high-coverage benchmarks, TASTE enables continuous, scalable evaluation of future agents.

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 execution and logging protocol. BenchAgent evaluates these substrate-internal workflows across ten reasoning, coding, and tool-use benchmarks with GPT-4.1, and separately reports a Protocol-Aligned External (PAE) GAIA study of a runtime-generated workflow. Under SI conditions, at most one of six tested MAS exceeds the matched single-agent anchor on benchmark-balanced average accuracy: EvoAgent lies within the Wilson one-run guidance, while the remaining five trail by 2.56-11.29 points and occupy more expensive accuracy-cost trade-offs. On the PAE GAIA snapshot, a Claude-Code-style runtime workflow reaches 66.72% overall and 69.23% on Level 3, more than 20 points above the strongest non-Claude baseline, Jarvis, a fixed MAS.

Modern agentic AI systems combine multiple large language model agents with heterogeneous skills, yet most architectures either fix communication in advance or allow full broadcast. Both can be inefficient because token cost, latency, redundancy, and error propagation increase with the number of active agents and communication links. We model agent selection and communication as a cooperative game with task-conditioned net utility U(Cmid x)=V(Cmid x)-sum_{iin C}c_i, separating coalition-level costs from agent activation costs. We propose a marginal-value activation rule and greedy router, extend the model to optimize communication edges with per-edge costs, and use estimated Shapley values to predict which agents are worth contacting before and during execution. We connect the problem to submodular maximization and prove two limited guarantees: a curvature-refined bound for a monotone, cardinality-constrained special case, and a tight 1/2-approximation, with a correction for signed objectives, for an unconstrained non-monotone case via double greedy. Neither guarantee applies directly to the main router, which remains a heuristic. We also prove a Shapley-submodularity sandwich bound linking the error of marginal-value routing to a per-agent diminishing-returns quantity. In synthetic experiments, greedy routing achieves 99.5% of brute-force-optimal utility while activating 1.96 of 8 agents on average, compared with 38.8% for full broadcast. Performance is robust to activation cost and redundancy weight but falls to 66% under strong violations of submodularity or noisy value estimates. We distinguish the framework from Shapley pricing, hedonic coalition formation, and communication-graph pruning, and propose evaluation on real multi-agent LLM benchmarks.

Mobile agent benchmarks increasingly rely on LLM-based judges to evaluate task completion, yet the reliability of these judges on mobile agent trajectories remains largely unexamined. We introduce MobileJudgeBench, a benchmark for systematically evaluating LLM-as-judge methods on mobile agent trajectories. Our benchmark comprises 931 human-annotated trajectories spanning 6 mobile agent benchmarks, 4 agent models, and 68 apps. We evaluate 6 judge methods (five adapted from SPA-Bench, A3 with two modes, AndroidArena, and AgentRewardBench, plus a simple baseline we design) across multiple LLM backends. Our experiments reveal three key findings. First, a simple baseline judge with sampled screenshots is competitive with, and often exceeds, purpose-built methods, indicating that more elaborate judge pipelines do not consistently improve judge quality; among competitive methods, the LLM backbone is the primary driver. Second, benchmark quality metrics reliably predict real-world judge utility: they correlate with both agent ranking fidelity for evaluation and downstream performance when judges serve as reward signals for on-policy reinforcement learning. Third, failure analysis across two LLM backends uncovers qualitatively opposite failure profiles, one conservative and the other permissive, linked to the backbone's precision-recall characteristics.

Recent advancements have underscored the potential of large language model (LLM)-based agents in financial decision-making. Despite this progress, the field currently encounters two main challenges: (1) the lack of a comprehensive LLM agent framework adaptable to a variety of financial tasks, and (2) the absence of standardized benchmarks and consistent datasets for assessing agent performance. To tackle these issues, we introduce InvestorBench, the first benchmark specifically designed for evaluating LLM-based agents in diverse financial decision-making contexts. InvestorBench enhances the versatility of LLM-enabled agents by providing a comprehensive suite of tasks applicable to different financial products, including single equities like stocks, cryptocurrencies and exchange-traded funds (ETFs). Additionally, we assess the reasoning and decision-making capabilities of our agent framework using thirteen different LLMs as backbone models, across various market environments and tasks. Furthermore, we have curated a diverse collection of open-source, multi-modal datasets and developed a comprehensive suite of environments for financial decision-making. This establishes a highly accessible platform for evaluating financial agents' performance across various scenarios.

Concepts serve as fundamental abstractions that support human reasoning and categorization. However, it remains unclear whether large language models truly capture such conceptual structures or primarily rely on surface-level pattern memorization. Existing benchmarks are largely static and fact oriented, which limits their ability to probe fine-grained semantic understanding and makes them vulnerable to data leakage and overfitting. To address this limitation, we introduce CK-Arena, a dynamic benchmark for conceptual knowledge evaluation based on a multi agent social deduction game, namely the Undercover game. In this setting, LLM based agents are assigned subtly different concept words and must describe, distinguish, and infer conceptual properties from others' statements. Model performance is evaluated through both game level outcomes and the semantic quality of generated descriptions. Furthermore, CK-Arena leverages the interaction process to automatically construct high quality question answering data for fine grained diagnostic analysis. Experimental results show that conceptual understanding varies substantially across models and categories, and is not strictly aligned with overall model capability. The data and code are available at the project homepage: https://ck-arena.site.

Large Language Models (LLMs) are increasingly serving as autonomous agents, and their utilization of external tools via the Model Context Protocol (MCP) is considered a future trend. Current MCP evaluation sets suffer from issues such as reliance on external MCP services and a lack of difficulty awareness. To address these limitations, we propose MCPAgentBench, a benchmark based on real-world MCP definitions designed to evaluate the tool-use capabilities of agents. We construct a dataset containing authentic tasks and simulated MCP tools. The evaluation employs a dynamic sandbox environment that presents agents with candidate tool lists containing distractors, thereby testing their tool selection and discrimination abilities. Furthermore, we introduce comprehensive metrics to measure both task completion rates and execution efficiency. Experiments conducted on various latest mainstream Large Language Models reveal significant performance differences in handling complex, multi-step tool invocations. All code is open-source at Github.

Large language models (LLMs) have significantly improved their ability to perform tasks in the field of code generation. However, there is still a gap between LLMs being capable coders and being top-tier software engineers. Based on the observation that top-level software engineers often ask clarifying questions to reduce ambiguity in both requirements and coding solutions, we argue that the same should be applied to LLMs for code generation tasks. In this work, we conducted an empirical study on the benchmark and analysis of the communication skills of LLMs for code generation. We define communication skills of LLMs as ``being able to ask clarifying questions when the description of the code generation problem has issues''. We created a new benchmark, HumanEvalComm, by modifying problem descriptions according to three issues: inconsistency, ambiguity, incompleteness. We defined new evaluation metrics such as Communication Rate and Good Question Rate, and then experimented on HumanEvalComm with different Code LLMs, and a new LLM agent approach, Okanagan, to identify and ask questions in ambiguous parts from code and descriptions for further refining the generated code. Finally, we discussed evaluation results by comparing Code LLMs and Okanagan with our findings.

LLM-based agents process external content, exposing them to prompt injection and multi-turn manipulation. Most safety benchmarks evaluate defenders against fixed attack pools collected before evaluation, single-turn or multi-turn. We present a 21-scenario benchmark for adaptive multi-round attacks against memoryless LLM defenders: an autonomous LLM attacker observes prior defender responses and pivots across rounds, while each defender response is evaluated as a fresh interaction. Holding the 21 scenarios, attackers, defenders, and structured-output scoring fixed, restricting scoring to the first attacker turn yields 0-1% attack success rate (ASR); allowing 15 rounds of adaptive attack yields 5.4-14.0%. Pooling three frontier attacker LLMs uncovers 1.4-2.2times as many unique successful attacks as the best single attacker, and the generated attacks have low cosine similarity (0.02-0.14) to attacks in existing benchmarks. Claude Opus 4.6 and GPT-5.4 are tied in aggregate (5.4% each; overlapping 95% CIs), but their weaknesses differ sharply: on one scenario Opus reaches 60% ASR (95% CI 36--80%) while GPT-5.4 and Gemini each stay at 7% (CI 1-30%; the gap is preserved in a higher-N replication). 13 of 21 scenarios distinguish at least one defender pair, yet rankings disagree across scenarios (Kendall's W = 0.19). We release the benchmark -- 21 evaluation scenarios, 10 public development scenarios, the orchestrator, baseline harnesses, and a multi-attacker CLI -- plus 945 transcripts from the 3times3 frontier matrix, an attack-replay dataset, and 18{,}422 gpt-oss-20b battles from an open competition's final scoring rounds.

The rise of LLM-based agents has opened new frontiers in AI applications, yet evaluating these agents remains a complex and underdeveloped area. This survey provides an in-depth overview of the emerging field of LLM agent evaluation, introducing a two-dimensional taxonomy that organizes existing work along (1) evaluation objectives -- what to evaluate, such as agent behavior, capabilities, reliability, and safety -- and (2) evaluation process -- how to evaluate, including interaction modes, datasets and benchmarks, metric computation methods, and tooling. In addition to taxonomy, we highlight enterprise-specific challenges, such as role-based access to data, the need for reliability guarantees, dynamic and long-horizon interactions, and compliance, which are often overlooked in current research. We also identify future research directions, including holistic, more realistic, and scalable evaluation. This work aims to bring clarity to the fragmented landscape of agent evaluation and provide a framework for systematic assessment, enabling researchers and practitioners to evaluate LLM agents for real-world deployment.

Agent benchmarks score submissions with outcome verifiers that are typically hand-written and brittle, leaving them open to reward hacking. We audit 1,968 tasks across five terminal-agent benchmarks and find 323 (16%) hackable by frontier models given only the task description. This corrupts both leaderboard rankings and RL training signal, yet the standard response is manual and reactive. We introduce the hacker-fixer loop, a method for building exploit-resistant verifiers without per-task manual patching. The loop alternates three LLM agents: a hacker tries to pass the verifier without solving the task, a fixer patches the verifier to reject each discovered exploit, and a solver confirms the patched verifier still admits legitimate solutions. The loop iterates: each patch reshapes what the verifier rewards, surfacing the next exploit. We further add verifier access, and let patches transfer across tasks, to broaden the exploits the loop discovers. On KernelBench, the loop drives the attack success rate from 62% to 0% on a held-out corpus of publicly reported exploits. We also find that weaker agents in the loop can defend against much stronger hackers: Gemini 3 Flash's loop drives the stronger Gemini 3.1 Pro and Claude Opus 4.7's attack success rate from 76% and 61% to 0% on KernelBench, and Gemini 3.1 Pro's from 39% to 17% on Terminal Bench across 77 tasks. We release Terminal Wrench (323 hackable environments, 3,632 hack trajectories) as a snapshot of the current attack surface, our patched verifiers, the exploits the loop discovered, and our implementation as a basis for future work.

We introduce HackSynth, a novel Large Language Model (LLM)-based agent capable of autonomous penetration testing. HackSynth's dual-module architecture includes a Planner and a Summarizer, which enable it to generate commands and process feedback iteratively. To benchmark HackSynth, we propose two new Capture The Flag (CTF)-based benchmark sets utilizing the popular platforms PicoCTF and OverTheWire. These benchmarks include two hundred challenges across diverse domains and difficulties, providing a standardized framework for evaluating LLM-based penetration testing agents. Based on these benchmarks, extensive experiments are presented, analyzing the core parameters of HackSynth, including creativity (temperature and top-p) and token utilization. Multiple open source and proprietary LLMs were used to measure the agent's capabilities. The experiments show that the agent performed best with the GPT-4o model, better than what the GPT-4o's system card suggests. We also discuss the safety and predictability of HackSynth's actions. Our findings indicate the potential of LLM-based agents in advancing autonomous penetration testing and the importance of robust safeguards. HackSynth and the benchmarks are publicly available to foster research on autonomous cybersecurity solutions.

Artificial Intelligence (AI) technology has emerged as a transformative force in financial analysis and the finance industry, though significant questions remain about the full capabilities of Large Language Model (LLM) agents in this domain. We present the Finance Agent Benchmark, featuring challenging and diverse real-world finance research problems that require LLMs to perform complex analysis using recent SEC filings. We construct the benchmark using a taxonomy of nine financial task categories, developed in consultation with experts from banks, hedge funds, and private equity firms. The dataset includes 537 expert-authored questions covering tasks from information retrieval to complex financial modeling, each validated through a rigorous review process to ensure accuracy and relevance. Moreover, we implement an agentic harness that equips LLMs with tools sufficient to produce accurate responses, including Google Search and EDGAR database access. Overall, the Finance Agent Benchmark provides a comprehensive testbed for measuring the progress of LLM-driven finance agents. Our evaluation reveals significant limitations in current AI capabilities - even the best-performing model (OpenAI o3) achieved only 46.8% accuracy at an average cost of $3.79 per query. This underscores the need for further advancements before reliable deployment in high-stakes finance settings.

The LLM Agent, equipped with a code interpreter, is capable of automatically solving real-world coding tasks, such as data analysis and image editing. However, existing benchmarks primarily focus on either simplistic tasks, such as completing a few lines of code, or on extremely complex and specific tasks at the repository level, neither of which are representative of various daily coding tasks. To address this gap, we introduce PyBench, a benchmark encompassing five main categories of real-world tasks, covering more than 10 types of files. Given a high-level user query and related files, the LLM Agent needs to reason and execute Python code via a code interpreter for a few turns before making a formal response to fulfill the user's requirements. Successfully addressing tasks in PyBench demands a robust understanding of various Python packages, superior reasoning capabilities, and the ability to incorporate feedback from executed code. Our evaluations indicate that current open-source LLMs are struggling with these tasks. Hence, we conduct analysis and experiments on four kinds of datasets proving that comprehensive abilities are needed for PyBench. Our fine-tuned 8B size model: PyLlama3 achieves an exciting performance on PyBench which surpasses many 33B and 70B size models. Our Benchmark, Training Dataset, and Model are available at: https://github.com/Mercury7353/PyBench{https://github.com/Mercury7353/PyBench}

The robustness of LLMs to jailbreak attacks, where users design prompts to circumvent safety measures and misuse model capabilities, has been studied primarily for LLMs acting as simple chatbots. Meanwhile, LLM agents -- which use external tools and can execute multi-stage tasks -- may pose a greater risk if misused, but their robustness remains underexplored. To facilitate research on LLM agent misuse, we propose a new benchmark called AgentHarm. The benchmark includes a diverse set of 110 explicitly malicious agent tasks (440 with augmentations), covering 11 harm categories including fraud, cybercrime, and harassment. In addition to measuring whether models refuse harmful agentic requests, scoring well on AgentHarm requires jailbroken agents to maintain their capabilities following an attack to complete a multi-step task. We evaluate a range of leading LLMs, and find (1) leading LLMs are surprisingly compliant with malicious agent requests without jailbreaking, (2) simple universal jailbreak templates can be adapted to effectively jailbreak agents, and (3) these jailbreaks enable coherent and malicious multi-step agent behavior and retain model capabilities. We publicly release AgentHarm to enable simple and reliable evaluation of attacks and defenses for LLM-based agents. We publicly release the benchmark at https://huggingface.co/ai-safety-institute/AgentHarm.

We introduce DABstep, a novel benchmark for evaluating AI agents on realistic multi-step data analysis tasks. DABstep comprises over 450 real-world challenges derived from a financial analytics platform, requiring models to combine code-based data processing with contextual reasoning over heterogeneous documentation. Each task demands an iterative, multi-step problem-solving approach, testing capabilities in data manipulation, cross-referencing multiple sources, and precise result reporting. The benchmark provides a factoid-style answer format with automatic correctness checks for objective scoring at scale. We evaluate leading LLM-based agents, revealing a substantial performance gap: even the best agent achieves only 14.55% accuracy on the hardest tasks. We detail our benchmark's design, dataset composition, task formulation, evaluation protocol, report baseline results and analyze failure modes. DABstep is released with a public leaderboard and toolkit to accelerate research in autonomous data analysis.

LLM agents are increasingly deployed in long-horizon, complex environments to solve challenging problems, but this expansion exposes them to long-horizon attacks that exploit multi-turn user-agent-environment interactions to achieve objectives infeasible in single-turn settings. To measure agent vulnerabilities to such risks, we present AgentLAB, the first benchmark dedicated to evaluating LLM agent susceptibility to adaptive, long-horizon attacks. Currently, AgentLAB supports five novel attack types including intent hijacking, tool chaining, task injection, objective drifting, and memory poisoning, spanning 28 realistic agentic environments, and 644 security test cases. Leveraging AgentLAB, we evaluate representative LLM agents and find that they remain highly susceptible to long-horizon attacks; moreover, defenses designed for single-turn interactions fail to reliably mitigate long-horizon threats. We anticipate that AgentLAB will serve as a valuable benchmark for tracking progress on securing LLM agents in practical settings. The benchmark is publicly available at https://tanqiujiang.github.io/AgentLAB_main.

LLM agents are expected to complete end-to-end units of work across software tools, business services, and local workspaces. Yet many agent benchmarks freeze a curated task set at release time and grade mainly the final response, making it difficult to evaluate agents against evolving workflow demand or verify whether a task was executed. We introduce Claw-Eval-Live, a live benchmark for workflow agents that separates a refreshable signal layer, updated across releases from public workflow-demand signals, from a reproducible, time-stamped release snapshot. Each release is constructed from public workflow-demand signals, with ClawHub Top-500 skills used in the current release, and materialized as controlled tasks with fixed fixtures, services, workspaces, and graders. For grading, Claw-Eval-Live records execution traces, audit logs, service state, and post-run workspace artifacts, using deterministic checks when evidence is sufficient and structured LLM judging only for semantic dimensions. The release contains 105 tasks spanning controlled business services and local workspace repair, and evaluates 13 frontier models under a shared public pass rule. Experiments reveal that reliable workflow automation remains far from solved: the leading model passes only 66.7% of tasks and no model reaches 70%. Failures are structured by task family and execution surface, with HR, management, and multi-system business workflows as persistent bottlenecks and local workspace repair comparatively easier but unsaturated. Leaderboard rank alone is insufficient because models with similar pass rates can diverge in overall completion, and task-level discrimination concentrates in a middle band of tasks. Claw-Eval-Live suggests that workflow-agent evaluation should be grounded twice, in fresh external demand and in verifiable agent action.

LLM agents are rapidly becoming the practical interface for task automation, yet the ecosystem lacks a principled way to choose among an exploding space of deployable configurations. Existing LLM leaderboards and tool/agent benchmarks evaluate components in isolation and remain fragmented across tasks, metrics, and candidate pools, leaving a critical research gap: there is little query-conditioned supervision for learning to recommend end-to-end agent configurations that couple a backbone model with a toolkit. We address this gap with AgentSelect, a benchmark that reframes agent selection as narrative query-to-agent recommendation over capability profiles and systematically converts heterogeneous evaluation artifacts into unified, positive-only interaction data. AgentSelectcomprises 111,179 queries, 107,721 deployable agents, and 251,103 interaction records aggregated from 40+ sources, spanning LLM-only, toolkit-only, and compositional agents. Our analyses reveal a regime shift from dense head reuse to long-tail, near one-off supervision, where popularity-based CF/GNN methods become fragile and content-aware capability matching is essential. We further show that Part~III synthesized compositional interactions are learnable, induce capability-sensitive behavior under controlled counterfactual edits, and improve coverage over realistic compositions; models trained on AgentSelect also transfer to a public agent marketplace (MuleRun), yielding consistent gains on an unseen catalog. Overall, AgentSelect provides the first unified data and evaluation infrastructure for agent recommendation, which establishes a reproducible foundation to study and accelerate the emerging agent ecosystem.

Recent research has leveraged large language model multi-agent systems for complex problem-solving while trying to reduce the manual effort required to build them, driving the development of automated agent workflow optimization methods. However, existing methods remain inflexible due to representational limitations, a lack of adaptability, and poor scalability when relying on discrete optimization techniques. We address these challenges with ScoreFlow, a simple yet high-performance framework that leverages efficient gradient-based optimization in a continuous space. ScoreFlow incorporates Score-DPO, a novel variant of the direct preference optimization method that accounts for quantitative feedback. Across six benchmarks spanning question answering, coding, and mathematical reasoning, ScoreFlow achieves an 8.2% improvement over existing baselines. Moreover, it empowers smaller models to outperform larger ones with lower inference costs. Project: https://github.com/Gen-Verse/ScoreFlow

With the rise of artificial intelligence (AI), applying large language models (LLMs) to Operations Research (OR) problem-solving has attracted increasing attention. Most existing approaches attempt to improve OR problem-solving through prompt engineering or fine-tuning strategies for LLMs. However, these methods are fundamentally constrained by the limited capabilities of non-reasoning LLMs. To overcome these limitations, we propose OR-LLM-Agent, an AI agent built on reasoning LLMs for automated OR problem solving. The agent decomposes the task into three sequential stages: mathematical modeling, code generation, and debugging. Each task is handled by a dedicated sub-agent, which enables more targeted reasoning. We also construct BWOR, a high-quality dataset for evaluating LLM performance on OR tasks. Our analysis shows that existing benchmarks such as NL4OPT, MAMO, and IndustryOR suffer from certain issues, making them less suitable for reliably evaluating LLM performance. In contrast, BWOR provides a more consistent and discriminative assessment of model capabilities. Experimental results demonstrate that OR-LLM-Agent outperforms advanced methods, including GPT-o3, Gemini 2.5 Pro, and ORLM, by at least 7% in accuracy. These results demonstrate the effectiveness of task decomposition for OR problem solving.

Deploying Large Language Model-based agents (LLM agents) in the public sector requires assuring that they meet the stringent legal, procedural, and structural requirements of public-sector institutions. Practitioners and researchers often turn to benchmarks for such assessments. However, it remains unclear what criteria benchmarks must meet to ensure they adequately reflect public-sector requirements, or how many existing benchmarks do so. In this paper, we first define such criteria based on a first-principles survey of public administration literature: benchmarks must be process-based, realistic, public-sector-specific and report metrics that reflect the unique requirements of the public sector. We analyse more than 1,300 benchmark papers for these criteria using an expert-validated LLM-assisted pipeline. Our results show that no single benchmark meets all of the criteria. Our findings provide a call to action for both researchers to develop public sector-relevant benchmarks and for public-sector officials to apply these criteria when evaluating their own agentic use cases.

Cutscenes are carefully choreographed cinematic sequences embedded in video games and interactive media, serving as the primary vehicle for narrative delivery, character development, and emotional engagement. Producing cutscenes is inherently complex: it demands seamless coordination across screenwriting, cinematography, character animation, voice acting, and technical direction, often requiring days to weeks of collaborative effort from multidisciplinary teams to produce minutes of polished content. In this work, we present Cutscene Agent, an LLM agent framework for automated end-to-end cutscene generation. The framework makes three contributions: (1)~a Cutscene Toolkit built on the Model Context Protocol (MCP) that establishes bidirectional integration between LLM agents and the game engine -- agents not only invoke engine operations but continuously observe real-time scene state, enabling closed-loop generation of editable engine-native cinematic assets; (2)~a multi-agent system where a director agent orchestrates specialist subagents for animation, cinematography, and sound design, augmented by a visual reasoning feedback loop for perception-driven refinement; and (3)~CutsceneBench, a hierarchical evaluation benchmark for cutscene generation. Unlike typical tool-use benchmarks that evaluate short, isolated function calls, cutscene generation requires long-horizon, multi-step orchestration of dozens of interdependent tool invocations with strict ordering constraints -- a capability dimension that existing benchmarks do not cover. We evaluate a range of LLMs on CutsceneBench and analyze their performance across this challenging task.

Large Language Model agents are increasingly augmented with agent skills. Current evaluation methods for skills remain limited. Most deployed benchmarks report only pass rate before and after a skill is attached, treating the skill as a black box change to agent behavior. We introduce Counterfactual Trace Auditing (CTA), a framework for measuring how a skill changes agent behavior. CTA pairs each with skill agent trace with a without skill counterpart on the same task, segments both traces into goal directed phases, aligns the phases, and emits structured Skill Influence Pattern (SIP) annotations. These annotations describe the behavioral effect of a skill rather than only its task outcome. We instantiate CTA on SWE-Skills-Bench with Claude across 49 software engineering tasks. The resulting audit reveals a clear evaluation gap. Pass rate changes by only +0.3 percentage points on average, suggesting little aggregate effect. Yet CTA identifies 522 SIP instances across the same paired traces, showing that the skills substantially reshape agent behavior even when pass rate is nearly unchanged. The audit also separates several recurring effects that pass rate cannot detect, including literal template copying, off task artifact creation, excess planning, and task recovery. Three findings emerge. First, high baseline tasks contain most of the observed skill effects, although their pass rate is already saturated and therefore cannot reflect those effects. Second, tasks with moderate baseline performance show the most recoverable gain, but often at substantially higher token cost. Third, the dominant SIP type can be identified by baseline bucket: surface anchoring is most common on ceiling tasks and edge-case prompting is most common on mid-range and floor tasks. These regularities turn informal failure mode observations into reproducible behavioral measurements.

Autonomous data analysis agents are increasingly expected to conduct exploratory analysis over underexplored data environments. This burden is especially salient in complex financial analytics, where relevant evidence is rarely pre-specified. However, existing benchmarks typically evaluate such agents in prior-guided settings, providing selected data sources, explicit data schemas, or cleaned data, thereby understating the exploratory burden. We introduce DataClawBench, a benchmark for exploratory real-world financial data analysis under limited prior guidance. DataClawBench contains approximately 2.06 million real-world records across enterprise, industry, and policy domains, with native data noise preserved. It further includes 492 cross-domain tasks derived from think-tank consulting scenarios, each annotated with intermediate milestones that diagnose exploration and reasoning failures beyond outcome accuracy. A systematic evaluation of eight advanced LLMs under the OpenClaw agent reveals that exploratory data analysis breaks agent reliability: more exploration does not reliably translate into task-relevant progress or correct final answers.

Parameterizing high-fidelity "digital twins" of batteries is a critical yet challenging inverse problem that hinders the pace of battery innovation. Prevailing methods formulate this as a black-box optimization (BBO) task, employing algorithms that are sample-inefficient and blind to the underlying physics. In this work, we introduce a new paradigm that reframes the inverse problem as a reasoning task, and present Battery-Sim-Agent, the first framework to deploy a Large Language Model (LLM) agent in a closed loop with a high-fidelity battery simulator. The agent mimics a human scientist's workflow: it interprets rich, multi-modal feedback from the simulator, forms physically-grounded hypotheses to explain discrepancies, and proposes structured parameter updates. On a systematically constructed benchmark suite spanning diverse battery chemistries, operating conditions, and difficulty levels, our agent significantly outperforms strong BBO baselines like Bayesian optimization in identifying accurate parameters. We further demonstrate the framework's capability in complex long-horizon degradation fitting tasks and validate its practical applicability on real-world battery datasets. Our results highlight the promise of LLM-agents as reasoning-based optimizers for scientific discovery and battery parameter estimation.

Large language model (LLM) agents increasingly execute multi-step workflows through tool use and interaction with users and environments. However, current agent evaluations are largely English-centric, limiting our understanding of agent capabilities in multilingual settings. We introduce BabelFlow, a benchmark-general agentic workflow that adapts existing agent benchmarks to new languages by analyzing runtime dependencies, coordinating structure-preserving translation, and combining multi-layer verification with human review to preserve task and evaluation semantics. Using BabelFlow, we construct BabelArena, a task-aligned benchmark comprising 16,146 instances derived from 702 canonical tasks across four benchmark families, 13 domains, and 23 languages. Experiments with five frontier models show that no single model dominates across benchmark families and that cross-language disparities extend well beyond task success. Lower-resource languages exhibit distinct failure patterns, with larger shares of tool-use and control-flow errors rather than answer-quality errors alone, pointing to gaps in reliable task execution across the resource levels of these languages. On the same tasks, agents in low-resource languages also consume substantially more tokens than in English (up to roughly twice the input) without proportional increases in interaction length, and language consistency degrades further on tasks requiring structured output, where switches are directed overwhelmingly toward English. We believe BabelArena provides a foundation for advancing research on reliable and efficient multilingual agents.

Large language model (LLM) agents have demonstrated strong capabilities across diverse domains. However, designing high-performing agentic systems remains challenging. Existing agent search methods suffer from three major limitations: (1) an emphasis on optimizing agentic workflows while under-utilizing proven human-designed components such as memory, planning, and tool use; (2) high evaluation costs, as each newly generated agent must be fully evaluated on benchmarks; and (3) inefficient search in large search space. In this work, we introduce a comprehensive framework to address these challenges. First, We propose a hierarchical search space that jointly models agentic workflow and composable functional components, enabling richer agentic system designs. Building on this structured design space, we introduce a predictive value model that estimates agent performance given agentic system and task description, allowing for efficient, low-cost evaluation during the search process. Finally, we present a hierarchical Monte Carlo Tree Search (MCTS) strategy informed by uncertainty to guide the search. Experiments on seven benchmarks, covering embodied, math, web, tool, and game, show that our method achieves an average performance gain of 8.34\% over state-of-the-art baselines and exhibits faster search progress with steeper improvement trajectories. Code repo is available at https://github.com/Ericccc02/AgentSwift.

Large language models (LLMs) are increasingly deployed as multi-step decision-making agents, where effective reward design is essential for guiding learning. Although recent work explores various forms of reward shaping and step-level credit assignment, a key signal remains largely overlooked: the intrinsic uncertainty of LLMs. Uncertainty reflects model confidence, reveals where exploration is needed, and offers valuable learning cues even in failed trajectories. We introduce SELAUR: Self Evolving LLM Agent via Uncertainty-aware Rewards, a reinforcement learning framework that incorporates uncertainty directly into the reward design. SELAUR integrates entropy-, least-confidence-, and margin-based metrics into a combined token-level uncertainty estimate, providing dense confidence-aligned supervision, and employs a failure-aware reward reshaping mechanism that injects these uncertainty signals into step- and trajectory-level rewards to improve exploration efficiency and learning stability. Experiments on two benchmarks, ALFWorld and WebShop, show that our method consistently improves success rates over strong baselines. Ablation studies further demonstrate how uncertainty signals enhance exploration and robustness.

Existing benchmarks for tool-using LLM agents primarily report single-run success rates and miss reliability properties required in production. We introduce ReliabilityBench, a benchmark for evaluating agent reliability across three dimensions: (i) consistency under repeated execution using pass^k, (ii) robustness to semantically equivalent task perturbations at intensity ε, and (iii) fault tolerance under controlled tool/API failures at intensity λ. ReliabilityBench contributes a unified reliability surface R(k,ε,λ), action metamorphic relations that define correctness via end-state equivalence rather than text similarity, and a chaos-engineering-style fault injection framework (timeouts, rate limits, partial responses, schema drift). We evaluate two models (Gemini 2.0 Flash, GPT-4o) and two agent architectures (ReAct, Reflexion) across four domains (scheduling, travel, customer support, e-commerce) over 1,280 episodes. Perturbations alone reduce success from 96.9% at ε=0 to 88.1% at ε=0.2. Rate limiting is the most damaging fault in ablations. ReAct is more robust than Reflexion under combined stress, and Gemini 2.0 Flash achieves comparable reliability to GPT-4o at much lower cost. ReliabilityBench provides a systematic framework for assessing production readiness of LLM agents.

Diagnosing and managing a patient is a complex, sequential decision making process that requires physicians to obtain information -- such as which tests to perform -- and to act upon it. Recent advances in artificial intelligence (AI) and large language models (LLMs) promise to profoundly impact clinical care. However, current evaluation schemes overrely on static medical question-answering benchmarks, falling short on interactive decision-making that is required in real-life clinical work. Here, we present AgentClinic: a multimodal benchmark to evaluate LLMs in their ability to operate as agents in simulated clinical environments. In our benchmark, the doctor agent must uncover the patient's diagnosis through dialogue and active data collection. We present two open medical agent benchmarks: a multimodal image and dialogue environment, AgentClinic-NEJM, and a dialogue-only environment, AgentClinic-MedQA. We embed cognitive and implicit biases both in patient and doctor agents to emulate realistic interactions between biased agents. We find that introducing bias leads to large reductions in diagnostic accuracy of the doctor agents, as well as reduced compliance, confidence, and follow-up consultation willingness in patient agents. Evaluating a suite of state-of-the-art LLMs, we find that several models that excel in benchmarks like MedQA are performing poorly in AgentClinic-MedQA. We find that the LLM used in the patient agent is an important factor for performance in the AgentClinic benchmark. We show that both having limited interactions as well as too many interaction reduces diagnostic accuracy in doctor agents. The code and data for this work is publicly available at https://AgentClinic.github.io.

Future prediction is a complex task for LLM agents, requiring a high level of analytical thinking, information gathering, contextual understanding, and decision-making under uncertainty. Agents must not only gather and interpret vast amounts of dynamic information but also integrate diverse data sources, weigh uncertainties, and adapt predictions based on emerging trends, just as human experts do in fields like politics, economics, and finance. Despite its importance, no large-scale benchmark exists for evaluating agents on future prediction, largely due to challenges in handling real-time updates and retrieving timely, accurate answers. To address this, we introduce FutureX, a dynamic and live evaluation benchmark specifically designed for LLM agents performing future prediction tasks. FutureX is the largest and most diverse live benchmark for future prediction, supporting real-time daily updates and eliminating data contamination through an automated pipeline for question gathering and answer collection. We evaluate 25 LLM/agent models, including those with reasoning, search capabilities, and integration of external tools such as the open-source Deep Research Agent and closed-source Deep Research models. This comprehensive evaluation assesses agents' adaptive reasoning and performance in dynamic environments. Additionally, we provide in-depth analyses of agents' failure modes and performance pitfalls in future-oriented tasks, including the vulnerability to fake web pages and the temporal validity. Our goal is to establish a dynamic, contamination-free evaluation standard that drives the development of LLM agents capable of performing at the level of professional human analysts in complex reasoning and predictive thinking.

In this paper, we introduce agentic unlearning which removes specified information from both model parameters and persistent memory in agents with closed-loop interaction. Existing unlearning methods target parameters alone, leaving two critical gaps: (i) parameter-memory backflow, where retrieval reactivates parametric remnants or memory artifacts reintroduce sensitive content, and (ii) the absence of a unified strategy that covers both parameter and memory pathways. We present Synchronized Backflow Unlearning (SBU), a framework that unlearns jointly across parameter and memory pathways. The memory pathway performs dependency closure-based unlearning that prunes isolated entities while logically invalidating shared artifacts. The parameter pathway employs stochastic reference alignment to guide model outputs toward a high-entropy prior. These pathways are integrated via a synchronized dual-update protocol, forming a closed-loop mechanism where memory unlearning and parametric suppression reinforce each other to prevent cross-pathway recontamination. Experiments on medical QA benchmarks show that SBU reduces traces of targeted private information across both pathways with limited degradation on retained data.

LLM-based agents are increasingly expected to handle real-world assistant tasks, yet existing benchmarks typically evaluate them under isolated sources of difficulty, such as a single environment or fully specified instructions. This leaves a substantial gap between current evaluation settings and the compositional challenges that arise in practical deployment. To address this gap, we introduce LiveClawBench, a benchmark to evaluate LLM agents on real-world assistant tasks. Based on an analysis of various real OpenClaw usage cases, we derive a Triple-Axis Complexity Framework that characterizes task difficulty along three dimensions: Environment Complexity, Cognitive Demand, and Runtime Adaptability. Guided by this framework, we construct a pilot benchmark with explicit complexity-factor annotations, covering real-world assistant tasks with compositional difficulty. Together, the framework and benchmark provide a principled foundation for evaluating LLM agents in realistic assistant settings, and establish a basis for future expansion across task domains and complexity axes. We are continuing to enrich our case collections to achieve more comprehensive domain and complexity coverage. The project page is at https://github.com/Mosi-AI/LiveClawBench.

Large language model (LLM) agents are increasingly expected to operate in enterprise environments, where work is distributed across specialized roles, permission-controlled systems, and cross-departmental procedures. However, existing enterprise benchmarks largely evaluate single agents with broad tool access, while existing multi-agent benchmarks rarely capture realistic enterprise constraints such as role specialization, access control, stateful business systems, and policy-based approvals. We introduce EntCollabBench, a benchmark for evaluating enterprise multi-agent collaboration. EntCollabBench simulates a permission-isolated organization with 11 role-specialized agents across six departments and contains two evaluation subsets: a Workflow subset, where agents collaboratively modify enterprise system states, and an Approval subset, where agents make policy-grounded decisions. Evaluation is based on execution traces, database state verification, and deterministic policy adjudication rather than natural-language response judging. Experiments with representative LLM agents show that current models still struggle with end-to-end enterprise collaboration, especially in delegation, context transfer, parameter grounding, workflow closure, and decision commitment. EntCollabBench provides a reproducible testbed for measuring and improving agent systems intended for realistic organizational environments.

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 about where execution broke down. We introduce CausalFlow, an interventional framework that converts failed agent traces into minimal counterfactual repairs and reusable supervision. CausalFlow models execution traces as sequential chains of dependent steps and computes Causal Responsibility Scores(CRS) via step-level counterfactual intervention to identify failure-inducing steps. For these steps, we generate minimally edited repairs that flip the final outcome to success, producing validated contrastive pairs of the form (wrong step, corrected step). CausalFlow supports two complementary uses: targeted test-time repair that recovers from failures with minimal behavioral drift, and training-time supervision suitable for offline preference optimization or reward modeling. Across four benchmarks spanning mathematical reasoning, code generation, question answering, and medical browsing, CausalFlow converts failed executions into validated minimal repairs with high minimality and causal-consensus scores, and demonstrates that causal attribution is necessary for reliable improvement across diverse agent tasks, outperforming heuristic refinement in complex retrieval settings while producing more localized repairs throughout. These results demonstrate that interventional analysis over structured execution traces provides a principled and scalable mechanism for transforming agent failures into reliability gains and learning-ready supervision.

Large language model (LLM) agents are increasingly deployed as teams, yet they fail in characteristically social ways: information holders stay silent, dominant speakers anchor groups on wrong answers, and norm violations go unsanctioned. We present Agent Sociodrama, a framework for modeling and validating evolving social dynamics in LLM agent teams. Under fixed personas, agents' emotional states and directed, asymmetric relationships are updated every round by a unified external LLM judge and shape their subsequent interactions. Across six LLM backbones and five social scenarios, we assess social fidelity through classical social-theory patterns and agreement with human judgments. The resulting trajectories exhibit structured social patterns and model-specific differences. We then introduce Trajectory-Guided Persona Recasting (TPR), which diagnoses group-dynamic failures from interaction transcripts and social trajectories and recasts agent personas before rerunning the task. Across four collaborative benchmarks, TPR improves performance from both heterogeneous and randomly initialized agent persona compositions, yielding up to 83.3%p of accuracy improvement over the initial casting. On representative distributed-information cooperation and mixed-motive bargaining tasks, we outperform SOTOPIA and AgentVerse in three of four comparisons, by up to 37.5%p in Pareto rate. These results show that validated social trajectories provide actionable signals for repairing LLM agent teams.

We introduce OneKE, a dockerized schema-guided knowledge extraction system, which can extract knowledge from the Web and raw PDF Books, and support various domains (science, news, etc.). Specifically, we design OneKE with multiple agents and a configure knowledge base. Different agents perform their respective roles, enabling support for various extraction scenarios. The configure knowledge base facilitates schema configuration, error case debugging and correction, further improving the performance. Empirical evaluations on benchmark datasets demonstrate OneKE's efficacy, while case studies further elucidate its adaptability to diverse tasks across multiple domains, highlighting its potential for broad applications. We have open-sourced the Code at https://github.com/zjunlp/OneKE and released a Video at http://oneke.openkg.cn/demo.mp4.

Integrating tools into Large Language Models (LLMs) has facilitated the widespread application. Despite this, in specialized downstream task contexts, reliance solely on tools is insufficient to fully address the complexities of the real world. This particularly restricts the effective deployment of LLMs in fields such as medicine. In this paper, we focus on the downstream tasks of medical calculators, which use standardized tests to assess an individual's health status. We introduce MeNTi, a universal agent architecture for LLMs. MeNTi integrates a specialized medical toolkit and employs meta-tool and nested calling mechanisms to enhance LLM tool utilization. Specifically, it achieves flexible tool selection and nested tool calling to address practical issues faced in intricate medical scenarios, including calculator selection, slot filling, and unit conversion. To assess the capabilities of LLMs for quantitative assessment throughout the clinical process of calculator scenarios, we introduce CalcQA. This benchmark requires LLMs to use medical calculators to perform calculations and assess patient health status. CalcQA is constructed by professional physicians and includes 100 case-calculator pairs, complemented by a toolkit of 281 medical tools. The experimental results demonstrate significant performance improvements with our framework. This research paves new directions for applying LLMs in demanding scenarios of medicine.

Code auditing is the process of reviewing code with the aim of identifying bugs. Large Language Models (LLMs) have demonstrated promising capabilities for this task without requiring compilation, while also supporting user-friendly customization. However, auditing a code repository with LLMs poses significant challenges: limited context windows and hallucinations can degrade the quality of bug reports, and analyzing large-scale repositories incurs substantial time and token costs, hindering efficiency and scalability. This work introduces an LLM-based agent, RepoAudit, designed to perform autonomous repository-level code auditing. Equipped with agent memory, RepoAudit explores the codebase on demand by analyzing data-flow facts along feasible program paths within individual functions. It further incorporates a validator module to mitigate hallucinations by verifying data-flow facts and checking the satisfiability of path conditions associated with potential bugs, thereby reducing false positives. RepoAudit detects 40 true bugs across 15 real-world benchmark projects with a precision of 78.43%, requiring on average only 0.44 hours and $2.54 per project. Also, it detects 185 new bugs in high-profile projects, among which 174 have been confirmed or fixed. We have open-sourced RepoAudit at https://github.com/PurCL/RepoAudit.

LLM-based coding agents rely on skills, pre-packaged instruction sets that extend agent capabilities, yet every token of skill content injected into the context window incurs both monetary cost and attention dilution. To understand the severity of this problem, we conduct a large-scale empirical study of 55,315 publicly available skills and find systemic inefficiencies: 26.4\% lack routing descriptions entirely, over 60\% of body content is non-actionable, and reference files can inject tens of thousands of tokens per invocation. Motivated by these findings, we present SkillReducer, a two-stage optimization framework. Stage~1 optimizes the routing layer by compressing verbose descriptions and generating missing ones via adversarial delta debugging. Stage~2 restructures skill bodies through taxonomy-driven classification and progressive disclosure, separating actionable core rules from supplementary content loaded on demand, validated by faithfulness checks and a self-correcting feedback loop. Evaluated on 600 skills and the SkillsBench benchmark, SkillReducer achieves 48\% description compression and 39\% body compression while improving functional quality by 2.8\%, revealing a less-is-more effect where removing non-essential content reduces distraction in the context window. These benefits transfer across five models from four families with a mean retention of 0.965, and generalize to an independent agent framework.

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 because mistakes can propagate across the trajectory: later actions may appear incorrect, but only because they depend on an earlier corrupted state. Therefore, failure attribution cannot be treated as independent step-level classification. We propose FALAT, a diagnostic framework for failure attribution in LLM agent trajectories. FALAT frames attribution as a dependency-guided search problem. It first constructs an expectation of how the task should be solved and uses this expectation to identify suspicious regions in the trajectory. It then traces dependencies among decisions, tool outputs, and agent messages to distinguish error-introducing steps from steps that merely inherit or propagate prior mistakes. Finally, FALAT evaluates whether correcting a candidate step would be sufficient to recover the expected outcome, allowing it to identify both the responsible agent and the decisive failure step. We evaluate FALAT on the Who&When benchmark, which includes both algorithm-generated and hand-crafted multi-agent failure trajectories. The results show that FALAT consistently improves responsible-agent and decisive-step attribution. Its best configurations achieve 46.0% step-level accuracy on algorithm-generated trajectories and 29.1% on the more challenging hand-crafted trajectories, outperforming specialized attribution baselines and direct prompting with standalone LLMs. These findings suggest that dependency-aware reasoning is essential for reliable failure diagnosis in LLM agent systems.

Large language models (LLMs) are expected to be trained to act as agents in various real-world environments, but this process relies on rich and varied tool-interaction sandboxes. However, access to real systems is often restricted; LLM-simulated environments are prone to hallucinations and inconsistencies; and manually built sandboxes are hard to scale. In this paper, we propose EnvScaler, an automated framework for scalable tool-interaction environments via programmatic synthesis. EnvScaler comprises two components. First, SkelBuilder constructs diverse environment skeletons through topic mining, logic modeling, and quality evaluation. Then, ScenGenerator generates multiple task scenarios and rule-based trajectory validation functions for each environment. With EnvScaler, we synthesize 191 environments and about 7K scenarios, and apply them to Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) for Qwen3 series models. Results on three benchmarks show that EnvScaler significantly improves LLMs' ability to solve tasks in complex environments involving multi-turn, multi-tool interactions. We release our code and data at https://github.com/RUC-NLPIR/EnvScaler.

Cumulative culture enables humans to preserve, reuse, and extend knowledge and skills across experiences and generations. Inspired by this principle, we introduce SimSkill, a self-evolving agent built around the Simulation of Urban MObility (SUMO) traffic simulator. SimSkill continually identifies capability gaps, generates and solves environment-grounded tasks, verifies solutions through an action--critic loop, and consolidates experience into episodic, procedural, and semantic memory. Through autonomous exploration, it builds a library of reusable skills and knowledge spanning major stages of the traffic-simulation workflow. We evaluate SimSkill on two held-out benchmarks across three backbone LLMs, with each result independently verified. It improves verified success by up to 25 percentage points, and ablations show complementary contributions from procedural and semantic memory. Its benefits remain backbone- and budget-dependent, as memory does not improve every model or uniformly reduce inference cost. More broadly, SimSkill illustrates a natural-language-centered design paradigm for LLM-based agent systems. Its high-level control logic, operating principles, and accumulated knowledge are expressed in natural language, while an LLM integrates them with executable tools and code to realize precise and reproducible execution. All code and experimental data are publicly available at https://github.com/qiliuchn/SimSkill-V1.

The evolution of Large Language Model (LLM) agents towards System~2 reasoning, characterized by deliberative, high-precision problem-solving, requires maintaining rigorous logical integrity over extended horizons. However, prevalent memory preprocessing paradigms suffer from destructive de-contextualization. By compressing complex sequential dependencies into pre-defined structures (e.g., embeddings or graphs), these methods sever the contextual integrity essential for deep reasoning. To address this, we propose E-mem, a framework shifting from Memory Preprocessing to Episodic Context Reconstruction. Inspired by biological engrams, E-mem employs a heterogeneous hierarchical architecture where multiple assistant agents maintain uncompressed memory contexts, while a central master agent orchestrates global planning. Unlike passive retrieval, our mechanism empowers assistants to locally reason within activated segments, extracting context-aware evidence before aggregation. Evaluations on the LoCoMo benchmark demonstrate that E-mem achieves over 54\% F1, surpassing the state-of-the-art GAM by 7.75\%, while reducing token cost by over 70\%.

Edge-cloud collaboration combines the lower inference overhead and greater data locality of edge-side execution with the stronger capabilities of cloud-hosted models. However, as LLM agents increasingly operate in long-horizon scenarios, cloud participation becomes interleaved throughout execution rather than a one-shot decision for each query. Such interleaved participation introduces trajectory-level effects that remain insufficiently understood. To address this gap, we introduce EC2-Bench, a benchmark for evaluating edge-cloud collaboration in long-horizon agent execution. It provides executable digital-workspace agent tasks and a unified evaluation framework covering task performance, resource usage, and privacy exposure. Using this benchmark, we empirically study diverse forms of edge-cloud collaboration. Our study reveals three findings. First, collaboration achieves a favorable balance among task performance, resource usage, and privacy exposure compared with edge-only and cloud-only execution. Second, as edge models become stronger, collaborative execution approaches cloud-only performance, and performance differences across strategies shrink, while privacy exposure remains strongly strategy-dependent. Third, stronger edge capability shifts preferred collaboration from externally controlled cloud participation and direct cloud execution toward more edge-controlled collaboration. Further analyses identify cloud-invocation decision-making and instruction following as important edge-side abilities for effective collaboration. These findings provide a systematic understanding of edge-cloud collaboration in long-horizon agent execution and insights for designing future edge agents that effectively collaborate with cloud models. Code and resources are available at https://github.com/OpenBMB/EC2-Bench.

Large language models (LLMs) have shown remarkable potential as human-like autonomous language agents to interact with real-world environments, especially for graphical user interface (GUI) automation. However, those GUI agents require comprehensive cognition ability including exhaustive perception and reliable action response. We propose Comprehensive Cognitive LLM Agent, CoCo-Agent, with two novel approaches, comprehensive environment perception (CEP) and conditional action prediction (CAP), to systematically improve the GUI automation performance. First, CEP facilitates the GUI perception through different aspects and granularity, including screenshots and complementary detailed layouts for the visual channel and historical actions for the textual channel. Second, CAP decomposes the action prediction into sub-problems: action type prediction and action target conditioned on the action type. With our technical design, our agent achieves new state-of-the-art performance on AITW and META-GUI benchmarks, showing promising abilities in realistic scenarios. Code is available at https://github.com/xbmxb/AAgent.

Complex agentic AI systems, powered by a coordinated ensemble of Large Language Models (LLMs), tool and memory modules, have demonstrated remarkable capabilities on intricate, multi-turn tasks. However, this success is shadowed by prohibitive economic costs and severe latency, exposing a critical, yet underexplored, trade-off. We formalize this challenge as the Agent System Trilemma: the inherent tension among achieving state-of-the-art performance, minimizing monetary cost, and ensuring rapid task completion. To dismantle this trilemma, we introduce EvoRoute, a self-evolving model routing paradigm that transcends static, pre-defined model assignments. Leveraging an ever-expanding knowledge base of prior experience, EvoRoute dynamically selects Pareto-optimal LLM backbones at each step, balancing accuracy, efficiency, and resource use, while continually refining its own selection policy through environment feedback. Experiments on challenging agentic benchmarks such as GAIA and BrowseComp+ demonstrate that EvoRoute, when integrated into off-the-shelf agentic systems, not only sustains or enhances system performance but also reduces execution cost by up to 80% and latency by over 70%.

Explaining observed phenomena through symbolic, interpretable formulas is a fundamental goal of science. Recently, large language models (LLMs) have emerged as promising tools for symbolic equation discovery, owing to their broad domain knowledge and strong reasoning capabilities. However, most existing LLM-based systems try to guess equations directly from data, without modeling the multi-step reasoning process that scientists often follow: first inferring physical properties such as symmetries, then using these as priors to restrict the space of candidate equations. We introduce KeplerAgent, an agentic framework that explicitly follows this scientific reasoning process. The agent coordinates physics-based tools to extract intermediate structure and uses these results to configure symbolic regression engines such as PySINDy and PySR, including their function libraries and structural constraints. Across a suite of physical equation benchmarks, KeplerAgent achieves substantially higher symbolic accuracy and greater robustness to noisy data than both LLM and traditional baselines.

LLM-powered tool-calling agents fulfill user requests by interacting with environments, querying data, and invoking tools in a multi-turn process. Yet, most existing benchmarks evaluate these systems under static environment interfaces, with fixed schemas and toolsets, making it difficult to assess how agents behave as environments evolves -- when capabilities are added, reorganized, or deprecated across successive environment versions. In this paper, we study structured environment evolution as a benchmark-construction problem for tool-calling agents. We propose ProEvolve, a graph-based framework that makes environment evolution programmable. At its core, a typed relational graph provides a unified, explicit representation of the environment - data, tools, and schema. Under this formalism, adding, removing, or modifying capabilities are expressed as graph transformations that coherently propagate updates across tools, schemas, and data access. Building on this, ProEvolve supports (1) automatic generation of evolved executable environments through explicit graph transformations, and (2) graph-grounded construction of task sandboxes via subgraph sampling and instantiation. We validate ProEvolve in two tool-calling domains, e-commerce and airline booking, in terms of quality, implementation validity, and failure modes. Finally, we use the generated benchmark as a downstream diagnostic to study how representative agents behave under structured environment evolution.

Running a business is a challenging form of intelligent work. Operators must infer opportunities from partial signals, commit capital under uncertainty, adapt to delayed outcomes in a changing market, and satisfy regulatory obligations before trading legally. Frontier LLM agents can increasingly complete complex workflows, yet business-related capabilities are rarely evaluated in existing agent benchmarks. We introduce Business Arena, a controlled environment where an AI agent runs a cross-border shop, buying from suppliers and selling to buyers over a long horizon. We ground the arena in real Alibaba.com sourcing data and market conditions calibrated from authoritative sources. Delayed and coupled consequences make individual business decisions difficult to judge, but their combined outcome is measurable through profit. Because profit alone cannot explain why an agent succeeds or fails, we compare agents with human-designed strategies to estimate available opportunity, use skill-level metrics to reveal underlying strengths and weaknesses, and trace realized gains and losses to the actions that produced them. We use mechanism ablations to establish that strong results reflect genuine business intelligence rather than neglect or simulator-specific shortcuts. We evaluate 15 frontier models and find a ninefold difference in mean final net worth. Even the best model falls behind human-designed strategies, indicating that business operation remains challenging for LLM agents. Skill-level analysis reveals operating styles, from margin-focused premium sellers to high-turnover wholesalers and customer-service specialists, while action-level attribution identifies the sourcing, pricing, and recovery decisions that create or destroy value. Together, Business Arena takes a first step toward a realistic and trustworthy testbed for evaluating end-to-end business agents.

Large language model (LLM) agents accomplish complex tasks through multi-turn interactions with external tools. However, malicious content embedded in user queries, tool descriptions, or environment observations can manipulate their decisions and induce unsafe actions. Adversarial co-evolution has emerged as an effective approach to strengthening LLM safety, but existing work primarily focuses on language-level safety or indirect prompt injection in web tasks. In this paper, we introduce AgentSafeEvo, a co-evolutionary reinforcement learning framework for agent safety built on an attacker-defender game spanning diverse tasks and attack types. The attacker learns to construct adversarial agent tasks that expose weaknesses in the defender's tool use, while the defender learns to complete tasks safely under these attacks. We further introduce learnability-guided task weighting, which uses within-task defender-reward variability to adjust each task's contribution to attacker and defender updates. Extensive experiments on agent safety and general capability benchmarks demonstrate that our method improves adversarial robustness while maintaining general agent capabilities.

Large language model (LLM) agents combine language-based reasoning with external tools to perform complex tasks. Adversarial inputs can exploit interactions between the agent and its environment, causing the agent to violate safety policies during execution. Yet existing evaluations often reduce agent safety to a single attack success rate (ASR), collapsing exposure, execution, observation, and adjudication and potentially conflating actual violations with evidence visibility. We introduce REDAgentBench, an executable framework for autonomous red-teaming and faithful measurement. It derives attacks from explicit safety constraints and associated agent-system vulnerabilities, runs them in isolated service sandboxes, and verifies harmful effects from service receipts and final-state changes. The benchmark contains 1,661 cases across five service surfaces. Across six models and three agent harnesses, macro-average ASR is 65.69%; reported ASR varies with harness and evidence view, while evaluation-context disclosure changes execution behavior. In a state-grounded diagnostic cohort, almost one in five confirmed violations with resolved action anchors occurs after the agent states the relevant constraint or risk, revealing a Recognition--Execution Gap. Finally, a training-free policy reminder reduces confirmed violations by more than 70 percentage points in matched replay. These findings show that executable evaluation can improve safety measurement and identify actionable intervention points.

Existing text-to-image models still struggle to generate images of multiple objects, especially in handling their spatial positions, relative sizes, overlapping, and attribute bindings. To efficiently address these challenges, we develop a training-free Multimodal-LLM agent (MuLan), as a human painter, that can progressively generate multi-object with intricate planning and feedback control. MuLan harnesses a large language model (LLM) to decompose a prompt to a sequence of sub-tasks, each generating only one object by stable diffusion, conditioned on previously generated objects. Unlike existing LLM-grounded methods, MuLan only produces a high-level plan at the beginning while the exact size and location of each object are determined upon each sub-task by an LLM and attention guidance. Moreover, MuLan adopts a vision-language model (VLM) to provide feedback to the image generated in each sub-task and control the diffusion model to re-generate the image if it violates the original prompt. Hence, each model in every step of MuLan only needs to address an easy sub-task it is specialized for. The multi-step process also allows human users to monitor the generation process and make preferred changes at any intermediate step via text prompts, thereby improving the human-AI collaboration experience. We collect 200 prompts containing multi-objects with spatial relationships and attribute bindings from different benchmarks to evaluate MuLan. The results demonstrate the superiority of MuLan in generating multiple objects over baselines and its creativity when collaborating with human users. The code is available at https://github.com/measure-infinity/mulan-code.

The rapid growth of biomedical data, tools, and literature has created a fragmented research landscape that outpaces human expertise. While AI agents offer a solution, they typically rely on static, manually curated toolsets, limiting their ability to adapt and scale. Here, we introduce STELLA, a self-evolving AI agent designed to overcome these limitations. STELLA employs a multi-agent architecture that autonomously improves its own capabilities through two core mechanisms: an evolving Template Library for reasoning strategies and a dynamic Tool Ocean that expands as a Tool Creation Agent automatically discovers and integrates new bioinformatics tools. This allows STELLA to learn from experience. We demonstrate that STELLA achieves state-of-the-art accuracy on a suite of biomedical benchmarks, scoring approximately 26\% on Humanity's Last Exam: Biomedicine, 54\% on LAB-Bench: DBQA, and 63\% on LAB-Bench: LitQA, outperforming leading models by up to 6 percentage points. More importantly, we show that its performance systematically improves with experience; for instance, its accuracy on the Humanity's Last Exam benchmark almost doubles with increased trials. STELLA represents a significant advance towards AI Agent systems that can learn and grow, dynamically scaling their expertise to accelerate the pace of biomedical discovery.

Large Language Model (LLM) agents are increasingly applied to complex, multi-step tasks that require interaction with diverse external tools across various domains. However, current LLM agent tool planning methods typically rely on greedy, reactive tool selection strategies that lack foresight and fail to account for inter-tool dependencies. In this paper, we present ToolTree, a novel Monte Carlo tree search-inspired planning paradigm for tool planning. ToolTree explores possible tool usage trajectories using a dual-stage LLM evaluation and bidirectional pruning mechanism that enables the agent to make informed, adaptive decisions over extended tool-use sequences while pruning less promising branches before and after the tool execution. Empirical evaluations across both open-set and closed-set tool planning tasks on 4 benchmarks demonstrate that ToolTree consistently improves performance while keeping the highest efficiency, achieving an average gain of around 10\% compared to the state-of-the-art planning paradigm.

Large language model (LLM) agents rely on external tools to solve complex tasks, but real-world toolsets often contain redundant tools with overlapping names and descriptions, introducing ambiguity and reducing selection accuracy. LLMs also face strict input context limits, preventing efficient consideration of large toolsets. To address these challenges, we propose ToolScope, which includes: (1) ToolScopeMerger with Auto-Correction to automatically audit and fix tool merges, reducing redundancy, and (2) ToolScopeRetriever to rank and select only the most relevant tools for each query, compressing toolsets to fit within context limits without sacrificing accuracy. Evaluations on three state-of-the-art LLMs and three open-source tool-use benchmarks show gains of 8.38% to 38.6% in tool selection accuracy, demonstrating ToolScope's effectiveness in enhancing LLM tool use.

Existing multi-agent benchmarks primarily test in competitive settings, short-horizon interactions under 20 steps, or simply aggregate individual performance, failing to isolate and highlight genuine collaboration capabilities of LLM-based agents. We introduce AgentWorld, a benchmark of 100 human-annotated tasks (with 100 augmented variants) for evaluating long-horizon, multi-agent collaboration. Tasks span 50+ interaction rounds across a rich MMORPG sandbox and require 3-20 agents with asymmetric roles and abilities to coordinate through communication, joint planning, and resource sharing under a blackbox setting where each agent acts independently without access to others' internal states. To quantify collaboration effectiveness in addition to conventional binary task success, we propose Causal Collaboration Effectiveness (CCE), a graph-based metric that traces causal dependencies between agent actions and measures what fraction of a team's effort actually contributed to the outcome. Experiments with Gemini 3 Flash, Claude Haiku 4.5, GPT-5 Mini, and DeepSeek R1-70B show that even the best model achieves only 52.0% task success, with systematic failure modes including communication breakdowns, role confusion, and inability to maintain shared plans across rounds. AgentWorld is fully open-source.

We introduce PaSa, an advanced Paper Search agent powered by large language models. PaSa can autonomously make a series of decisions, including invoking search tools, reading papers, and selecting relevant references, to ultimately obtain comprehensive and accurate results for complex scholarly queries. We optimize PaSa using reinforcement learning with a synthetic dataset, AutoScholarQuery, which includes 35k fine-grained academic queries and corresponding papers sourced from top-tier AI conference publications. Additionally, we develop RealScholarQuery, a benchmark collecting real-world academic queries to assess PaSa performance in more realistic scenarios. Despite being trained on synthetic data, PaSa significantly outperforms existing baselines on RealScholarQuery, including Google, Google Scholar, Google with GPT-4 for paraphrased queries, chatGPT (search-enabled GPT-4o), GPT-o1, and PaSa-GPT-4o (PaSa implemented by prompting GPT-4o). Notably, PaSa-7B surpasses the best Google-based baseline, Google with GPT-4o, by 37.78% in recall@20 and 39.90% in recall@50. It also exceeds PaSa-GPT-4o by 30.36% in recall and 4.25% in precision. Model, datasets, and code are available at https://github.com/bytedance/pasa.

Large language model (LLM) agents have demonstrated remarkable potential in advancing scientific discovery. However, their capability in the fundamental yet crucial task of reproducing code from research papers, especially in the NLP domain, remains underexplored. This task includes unique complex reasoning challenges in the intellectual synthesis of abstract concepts and the comprehension of code repositories with interdependent files. Motivated by this gap, we present LMR-BENCH, a benchmark designed to systematically evaluate the capability of LLM agents on code reproduction from Language Modeling Research. It consists of 28 code reproduction tasks derived from 23 research papers published in top-tier NLP venues over the past five years, spanning nine fundamental categories. Models are provided with a research paper, a code repository containing one or more masked functions, and instructions for implementing these functions. We conduct extensive experiments in standard prompting and LLM agent settings with state-of-the-art LLMs, evaluating the accuracy of unit tests and performing LLM-based evaluation of code correctness. Experimental results reveal that even the most advanced models still exhibit persistent limitations in scientific reasoning and code synthesis, highlighting critical gaps in LLM agents' ability to autonomously reproduce scientific research

Multimodal AI agents are increasingly automating complex real-world workflows that involve online web execution. However, current web-agent benchmarks suffer from a critical limitation: they focus entirely on web-based interaction and perception, lacking grounding in the user's real-world physical surroundings. This limitation prevents evaluation in crucial scenarios, such as when an agent must use egocentric visual perception (e.g., via AR glasses) to recognize an object in the user's surroundings and then complete a related task online. To address this gap, we introduce Ego2Web, the first benchmark designed to bridge egocentric video perception and web agent execution. Ego2Web pairs real-world first-person video recordings with web tasks that require visual understanding, web task planning, and interaction in an online environment for successful completion. We utilize an automatic data-generation pipeline combined with human verification and refinement to curate well-constructed, high-quality video-task pairs across diverse web task types, including e-commerce, media retrieval, knowledge lookup, etc. To facilitate accurate and scalable evaluation for our benchmark, we also develop a novel LLM-as-a-Judge automatic evaluation method, Ego2WebJudge, which achieves approximately 84% agreement with human judgment, substantially higher than existing evaluation methods. Experiments with diverse SoTA agents on our Ego2Web show that their performance is weak, with substantial headroom across all task categories. We also conduct a comprehensive ablation study on task design, highlighting the necessity of accurate video understanding in the proposed task and the limitations of current agents. We hope Ego2Web can be a critical new resource for developing truly capable AI assistants that can seamlessly see, understand, and act across the physical and digital worlds.

Scientific research relies on accurate information retrieval from literature to support analytical decisions. In this work, we introduce a new task, INformation reTRieval through literAture reVIEW (IntraView), which aims to automate fine-grained information retrieval faithfully grounded in the provided content in response to research-driven queries, and propose IntrAgent, an LLM-based agent that addresses this challenging task. In particular, IntrAgent is designed to mimic human behaviors when reading literature for information retrieval -- identifying relevant sections and then iteratively extracting key details to refine the retrieved information. It follows a two-stage pipeline: a Section Ranking stage that prioritizes relevant literature sections through structural-knowledge-enabled reasoning, and an Iterative Reading stage that continuously extracts details and synthesizes them into concise, contextually grounded answers. To support rigorous evaluation, we introduce IntraBench, a new benchmark consisting of 315 test instances built from expert-authored questions paired with literature spanning five STEM domains. Across seven backbone LLMs, IntrAgent achieves on average 13.2% higher cross-domain accuracy than state-of-the-art RAG and research-agent baselines.

Recent advances in LLM agentic systems have improved the automation of offensive security tasks, particularly for Capture the Flag (CTF) challenges. We systematically investigate the key factors that drive agent success and provide a detailed recipe for building effective LLM-based offensive security agents. First, we present CTFJudge, a framework leveraging LLM as a judge to analyze agent trajectories and provide granular evaluation across CTF solving steps. Second, we propose a novel metric, CTF Competency Index (CCI) for partial correctness, revealing how closely agent solutions align with human-crafted gold standards. Third, we examine how LLM hyperparameters, namely temperature, top-p, and maximum token length, influence agent performance and automated cybersecurity task planning. For rapid evaluation, we present CTFTiny, a curated benchmark of 50 representative CTF challenges across binary exploitation, web, reverse engineering, forensics, and cryptography. Our findings identify optimal multi-agent coordination settings and lay the groundwork for future LLM agent research in cybersecurity. We make CTFTiny open source to public https://github.com/NYU-LLM-CTF/CTFTiny along with CTFJudge on https://github.com/NYU-LLM-CTF/CTFJudge.

Digital Adoption Platforms (DAPs) are embedded overlays widely used on web systems to guide users through operations inside a page, helping them get started with unfamiliar interfaces quickly. Completing a real task, however, rarely means clicking a few buttons on a single page: it takes a sequence of actions that unfolds across changing page states. Prior studies have also treated automated web agent actions and guide text generation as two separate problems, and most of them feed models textual page representations such as the DOM or accessibility trees rather than the rendered screens that humans actually operate on. In this work we introduce MAG, the first benchmark that unifies task execution and guide writing into a single Multimodal Action and Guide task, with two grounding schemes over screenshots: Set-of-Mark element selection and raw pixel coordinates. We further build a complete harness for this compound task, covering annotation with LLM assistance and human verification, training, evaluation in live environments, and joint metrics for actions and guides. With this harness we evaluate frontier API models and open multimodal models, and report detailed analyses. Finally, we design a GRPO training method augmented with expert trajectories, which nearly doubles the success rate of a supervised 9B agent (from 6.9% to 13.2%) and improves guide quality at the same time. Even the strongest model completes fewer than 40% of the tasks, leaving ample room for future research.

Operational disaster response goes beyond damage assessment, requiring responders to integrate multi-sensor signals, reason over road networks, populations and key facilities, plan evacuations, and produce actionable reports. However, prior work largely isolates remote-sensing perception or evaluates generic tool use, leaving the end-to-end workflows of emergency operations underexplored. In this paper, we introduce Disaster Operational Response Agent benchmark (DORA), the first agentic benchmark for end-to-end disaster response: 515 expert-authored tasks across 45 real-world disaster events spanning 10 types, paired with expert-verified, replayable gold trajectories totaling 3,500 tool-call steps. Tasks span five dimensions that cover the operational disaster-response pipeline: disaster perception, spatial relational analysis, rescue and evacuation planning, temporal evolution reasoning, and multi-modal report synthesis. Agents compose calls from a 108-tool MCP library over heterogeneous geospatial data: optical, SAR, and multi-spectral imagery across single-, bi-, and multi-temporal sequences (0.015-10m GSD), complemented by elevation and social vector layers. We comprehensively evaluate 13 frontier LLMs on our benchmark, revealing three persistent challenges: 1) disaster-domain grounding exposes unique failure modes (damage-semantic grounding, sensor-modality mismatch, and disaster-pipeline composition); 2) agents are doubly bottlenecked by tool selection and argument grounding, where gold tool-order hints improve accuracy by only 1.08-4.40%, and alternative scaffolds yield at most a 3.24% gain; 3) compositional fragility scales with trajectory length, the agent-to-gold gap widening from 7% to 56% on long pipelines. DORA establishes a rigorous testbed for operationally reliable disaster-response agents.

Computation cost is a critical challenge for LLM agent systems, as their long-horizon trajectories continuously accumulate computational overhead. While existing studies focus on improving the overall efficiency of agent systems, the efficiency of individual steps remains largely unexplored. In this paper, we propose and formulate a new research problem: redundant step detection, aiming to detect these steps that consume substantial computational resources while contributing little or nothing to task completion. To support this initiative, we introduce RedundancyBench, a new benchmark that contains 1,920 trajectories collected from 6 different LLMs with 3,613 step-level annotations. Using RedundancyBench, we develop and evaluate 6 frontier LLMs and Jev as redundant step detectors, summarizing their strengths and limitations. Our experiments reveal that the redundant step detection is still challenging even for SOTA models, such as Claude-Opus-5, particularly without prior information. Code and dataset in this paper are both available in https://anonymous.4open.science/r/RedundancyBench-E0CB.

Evaluating agents by outcomes alone can obscure the capabilities that produce them. This problem is especially pronounced in evolving environments, where outcomes reflect a closed-loop interaction between agent behavior and changing external conditions. We introduce LiveMACEBench, a process-aware benchmark that uses live financial markets as a naturally evolving testbed for persistent LLM agents. Five frontier LLMs operate along continuous trajectories under matched Tool Use, Persistent Memory, Rule Following, and Multi-Agent Collaboration configurations. We evaluate them through both realized outcomes and mechanism-specific diagnostics derived from complete decision traces. Across 30 days of live evaluation, we find a pronounced outcome-capability gap: realized returns often diverge from capability-specific measurements, and similar outcomes can arise from markedly different patterns of mechanism use. Trace-level diagnostics further expose distinct bottlenecks across capabilities, demonstrating that mechanism access, effective mechanism use, and downstream performance are not interchangeable measures of agent capability. LiveMACEBench makes this distinction measurable, turning live markets from a performance leaderboard into a diagnostic environment for agent capability

As large language models (LLMs) acquire advanced biological design capabilities, they offer substantial research benefits but simultaneously introduce severe dual-use risks, such as engineering enhanced pathogens or drug-resistant bacteria. While biosecurity frameworks recognize these threats, the actual risk of LLMs executing dangerous biological modifications remains unquantified, particularly when augmented with specialized biological design tools (BDTs). Existing evaluations primarily assess hazardous knowledge possession rather than practical application, overlook risk amplification from BDTs, and rely on unreliable, expensive, or hazardous validation methods. To address these gaps, we introduce DRYLAB-Bench (Design Risk Yielded by LLM Agent in Biodesign), the first comprehensive benchmark evaluating the high-risk biological mutation design capabilities of LLM agents. Moving beyond basic question-answering, DRYLAB-Bench tasks agents with 30 genuinely hazardous mutation-design scenarios distilled from biosecurity guidelines. It evaluates models in a multi-round, tool-augmented environment where LLMs orchestrate BDTs to iteratively refine candidates. Crucially, it scores designs directly against 187,695 empirical wet-lab measurements from ProteinGym and ViroGym using our proposed Biosecurity Risk Score (BRS), quantifying risk-associated effects while enabling scalable and safe ground-truth verification. Evaluations of 15 frontier models reveal substantial baseline threats: without tools, 67.6% of one-shot designs exceed the average measured variant, and 21.4% reach the highest-risk decile. Equipping models with BDTs significantly amplifies this risk, raising the pooled risk by \(ΔBRS=+0.116\) and the top-decile share to 23.5%. Code: https://github.com/drylab-bio/DRYLAB-Bench

Large Language Model (LLM) agents significantly extend the capabilities of standalone LLMs, empowering them to interact with external tools (e.g., APIs, functions) and complete complex tasks in a self-directed fashion. The challenge of tool use demands that LLMs not only understand user queries and generate answers but also excel in task planning, memory management, tool invocation, and result summarization. While traditional approaches focus on training a single LLM with all these capabilities, performance limitations become apparent, particularly with smaller models. Moreover, the entire LLM may require retraining when tools are updated. To overcome these challenges, we propose a novel strategy that decomposes the aforementioned capabilities into a planner, caller, and summarizer. Each component is implemented by a single LLM that focuses on a specific capability and collaborates with other components to accomplish the task. This modular framework facilitates individual updates and the potential use of smaller LLMs for building each capability. To effectively train this framework, we introduce a two-stage training paradigm. First, we fine-tune a backbone LLM on the entire dataset without discriminating sub-tasks, providing the model with a comprehensive understanding of the task. Second, the fine-tuned LLM is used to instantiate the planner, caller, and summarizer respectively, which are continually fine-tuned on respective sub-tasks. Evaluation across various tool-use benchmarks illustrates that our proposed multi-LLM framework surpasses the traditional single-LLM approach, highlighting its efficacy and advantages in tool learning.

This paper presents a novel approach for unified retrieval-augmented generation (RAG) systems using the recent emerging large language model (LLM) agent concept. Specifically, Agent LLM, which utilizes LLM as fundamental controllers, has become a promising approach to enable the interpretability of RAG tasks, especially for complex reasoning question-answering systems (e.g., multi-hop queries). Nonetheless, previous works mainly focus on solving RAG systems with either single-hop or multi-hop approaches separately, which limits the application of those approaches to real-world applications. In this study, we propose a trainable agent framework called Agent-UniRAG for unified retrieval-augmented LLM systems, which enhances the effectiveness and interpretability of RAG systems. The main idea is to design an LLM agent framework to solve RAG tasks step-by-step based on the complexity of the inputs, simultaneously including single-hop and multi-hop queries in an end-to-end manner. Furthermore, we introduce SynAgent-RAG, a synthetic dataset to enable the proposed agent framework for small open-source LLMs (e.g., Llama-3-8B). The results show comparable performances with closed-source and larger open-source LLMs across various RAG benchmarks. Our source code and dataset are publicly available for further exploitation.

Smartphone agents are increasingly important for helping users control devices efficiently, with (Multimodal) Large Language Model (MLLM)-based approaches emerging as key contenders. Fairly comparing these agents is essential but challenging, requiring a varied task scope, the integration of agents with different implementations, and a generalisable evaluation pipeline to assess their strengths and weaknesses. In this paper, we present SPA-Bench, a comprehensive SmartPhone Agent Benchmark designed to evaluate (M)LLM-based agents in an interactive environment that simulates real-world conditions. SPA-Bench offers three key contributions: (1) A diverse set of tasks covering system and third-party apps in both English and Chinese, focusing on features commonly used in daily routines; (2) A plug-and-play framework enabling real-time agent interaction with Android devices, integrating over ten agents with the flexibility to add more; (3) A novel evaluation pipeline that automatically assesses agent performance across multiple dimensions, encompassing seven metrics related to task completion and resource consumption. Our extensive experiments across tasks and agents reveal challenges like interpreting mobile user interfaces, action grounding, memory retention, and execution costs. We propose future research directions to ease these difficulties, moving closer to real-world smartphone agent applications. SPA-Bench is available at https://ai-agents-2030.github.io/SPA-Bench/.

Recently, web-based Large Language Model (LLM) agents autonomously perform increasingly complex tasks, thereby bringing significant convenience. However, they also amplify the risks of malicious misuse cases such as unauthorized collection of personally identifiable information (PII), generation of socially divisive content, and even automated web hacking. To address these threats, we propose an AI Kill Switch technique that can immediately halt the operation of malicious web-based LLM agents. To achieve this, we introduce AutoGuard - the key idea is generating defensive prompts that trigger the safety mechanisms of malicious LLM agents. In particular, generated defense prompts are transparently embedded into the website's DOM so that they remain invisible to human users but can be detected by the crawling process of malicious agents, triggering its internal safety mechanisms to abort malicious actions once read. To evaluate our approach, we constructed a dedicated benchmark consisting of three representative malicious scenarios (PII collection, social rift content generation, and web hacking attempts). Experimental results show that the AutoGuard method achieves over 80% Defense Success Rate (DSR) on malicious agents, including GPT-4o, Claude-3, and Llama3.3-70B-Instruct. It also maintains strong performance, achieving around 90% DSR on GPT-5, GPT-4.1, and Gemini-2.5-Flash when used as the malicious agent, demonstrating robust generalization across models and scenarios. Through this research, we have demonstrated the controllability of web-based LLM agents across various scenarios and models, thereby contributing to the broader effort of AI control and safety.

Large language model (LLM) agents now execute long, tool-using tasks where final outcome checks can arrive too late for intervention. Online warning requires lightweight prefix monitors over heterogeneous traces, but hand-authored event schemas are brittle and deployment-time LLM judging is costly. We introduce PrefixGuard, a trace-to-monitor framework with an offline StepView induction step followed by supervised monitor training. StepView induces deterministic typed-step adapters from raw trace samples, and the monitor learns an event abstraction and prefix-risk scorer from terminal outcomes. Across WebArena, τ^2-Bench, SkillsBench, and TerminalBench, the strongest PrefixGuard monitors reach 0.900/0.710/0.533/0.557 AUPRC. Using the strongest backend within each representation, they improve over raw-text controls by an average of +0.137 AUPRC. LLM judges remain substantially weaker under the same prefix-warning protocol. We also derive an observability ceiling on score-based area under the precision-recall curve (AUPRC) that separates monitor error from failures lacking evidence in the observed prefix. For finite-state audit, post-hoc deterministic finite automaton (DFA) extraction remains compact on WebArena and τ^2-Bench (29 and 20 states) but expands to 151 and 187 states on SkillsBench and TerminalBench. Finally, first-alert diagnostics show that strong ranking does not imply deployment utility: WebArena ranks well yet fails to support low-false-alarm alerts, whereas τ^2-Bench and TerminalBench retain more actionable early alerts. Together, these results position PrefixGuard as a practical monitor-synthesis recipe with explicit diagnostics for when prefix warnings translate into actionable interventions.

Agent benchmarks are increasingly used to compare large language models (LLMs) and guide deployment decisions, yet benchmark scores are meaningful only if they measure model capability rather than properties of the evaluation pipeline. We identify a double measurement confound: execution-critical decisions are performed by a fixed scaffold instead of the model, while the scorer evaluates outputs using criteria that may not reflect task correctness. We unify these issues within a measurement-theoretic framework that characterizes when benchmark scores can be interpreted as evidence of model capability, and instantiate it with an audit-and-repair protocol that (i) transfers execution-critical decisions from the scaffold to the model, (ii) replaces shape-based evaluation with seeded ground-truth scoring, and (iii) reports reliability beyond the mean through worst-case and tail-risk metrics. Experiments on ComtradeBench show that the joint intervention transforms a nearly flat leaderboard into a reliability spectrum that distinguishes both average performance and robustness across seeds. Applying the audit to existing benchmarks further shows that scorer validity is benchmark-specific, whereas scaffold ownership is an uncontrolled axis wherever we probed it. Our results suggest that benchmark scores should be interpreted together with their scaffolding level, scoring criterion, and reliability profile, providing a practical framework for more valid evaluation of LLM agents.

Agents need reliable ways to validate their outputs, both when completing tasks and when filtering synthetic training data. We study verifier generation: from the instruction and environment available for solving a task, an agent writes an executable verifier that must accept every valid solution and reject invalid ones. We introduce VerifierBench, a benchmark of 222 tasks from six agent benchmarks spanning tool-call, state-based, and code verification, with 4,993 hidden test cases that include alternative valid solutions and attacks; Verifier4Verifier (V4V), a six-check audit, screens its source verifiers and test suites. Across 19 models in a common agent harness, normalized balanced accuracy ranges from 10.6 to 62.1, with generated verifiers consistently rejecting invalid solutions more reliably than they accept valid ones. This apparent strength can mask overly restrictive verification: most false rejections in the analyzed records come from verifiers that accept no tested solution to a task. Case studies reveal unsupported assumptions about solution formats, tool use, and code changes that exclude valid solutions. These errors survive self-testing because agents test solutions that share the verifier's mistaken assumptions, or submit verifiers without confirming that any correct solution passes. These findings suggest testing both whether a verifier accepts a correct solution and whether it accepts other solutions permitted by the task. The project page and dataset are publicly available.

This paper introduces OSC (Orchestrating Cognitive Synergy), a knowledge-aware adaptive collaboration framework designed to enhance cognitive synergy in multi-agent systems with large language models. While prior work has advanced agent selection and result aggregation, efficient linguistic interactions for deep collaboration among expert agents remain a critical bottleneck. OSC addresses this gap as a pivotal intermediate layer between selection and aggregation, introducing Collaborator Knowledge Models (CKM) to enable each agent to dynamically perceive its collaborators' cognitive states. Through real-time cognitive gap analysis, agents adaptively adjust communication behaviors, including content focus, detail level, and expression style, using learned strategies. Experiments on complex reasoning and problem-solving benchmarks demonstrate that OSC significantly improves task performance and communication efficiency, transforming "parallel-working individuals'' into a "deeply collaborative cognitive team.'' This framework not only optimizes multi-agent collaboration but also offers new insights into LLM agent interaction behaviors.

Multi-turn agent systems based on Large Language Models (LLMs) have been increasingly popular for software engineering tasks. While LLM agents show decent effectiveness, the high computational cost of input tokens due to the ever-growing trajectory remains an efficiency concern for their applications. Efficiency is largely neglected in existing studies and agent products, and this paper fills the gap by introducing an inference-time trajectory reduction approach to reduce the cost of agents. Through analyzing existing agent trajectories, we demonstrate that useless, redundant, and expired information is widespread in all trajectories, which can be identified and reduced without harming the agent's performance. We then design a simple yet effective trajectory reduction approach, AgentDiet, which automatically removes such waste information. We implement AgentDiet on a top-performing coding agent, and the evaluation on two LLMs and two benchmarks shows that AgentDiet can reduce input tokens by 39.9% ~ 59.7%, or the final computational cost by 21.1% ~ 35.9%, while maintaining the same agent performance. This indicates that trajectory reduction is a promising direction for agent systems.

This paper introduces KLong, an open-source LLM agent trained to solve extremely long-horizon tasks. The principle is to first cold-start the model via trajectory-splitting SFT, then scale it via progressive RL training. Specifically, we first activate basic agentic abilities of a base model with a comprehensive SFT recipe. Then, we introduce Research-Factory, an automated pipeline that generates high-quality training data by collecting research papers and constructing evaluation rubrics. Using this pipeline, we build thousands of long-horizon trajectories distilled from Claude 4.5 Sonnet (Thinking). To train with these extremely long trajectories, we propose a new trajectory-splitting SFT, which preserves early context, progressively truncates later context, and maintains overlap between sub-trajectories. In addition, to further improve long-horizon task-solving capability, we propose a novel progressive RL, which schedules training into multiple stages with progressively extended timeouts. Experiments demonstrate the superiority and generalization of KLong, as shown in Figure 1. Notably, our proposed KLong (106B) surpasses Kimi K2 Thinking (1T) by 11.28% on PaperBench, and the performance improvement generalizes to other coding benchmarks like SWE-bench Verified and MLE-bench.

Agent skills, which are reusable, domain-specific knowledge artifacts, have become a popular mechanism for extending LLM-based agents, yet formally benchmarking skill usage performance remains scarce. Existing skill benchmarking efforts focus on overly idealized conditions, where LLMs are directly provided with hand-crafted, narrowly-tailored task-specific skills for each task, whereas in many realistic settings, the LLM agent may have to search for and select relevant skills on its own, and even the closest matching skills may not be well-tailored for the task. In this paper, we conduct the first comprehensive study of skill utility under progressively challenging realistic settings, where agents must retrieve skills from a large collection of 34k real-world skills and may not have access to any hand-curated skills. Our findings reveal that the benefits of skills are fragile: performance gains degrade consistently as settings become more realistic, with pass rates approaching no-skill baselines in the most challenging scenarios. To narrow this gap, we study skill refinement strategies, including query-specific and query-agnostic approaches, and we show that query-specific refinement substantially recovers lost performance when the initial skills are of reasonable relevance and quality. We further demonstrate the generality of retrieval and refinement on Terminal-Bench 2.0, where they improve the pass rate of Claude Opus 4.6 from 57.7% to 65.5%. Our results, consistent across multiple models, highlight both the promise and the current limitations of skills for LLM-based agents. Our code is available at https://github.com/UCSB-NLP-Chang/Skill-Usage.

Large Language Model (LLM) agents have shown significant autonomous capabilities in dynamically searching and incorporating relevant tools or Model Context Protocol (MCP) servers for individual queries. However, fixed context windows limit effectiveness in multi-turn interactions requiring repeated, independent tool usage. We introduce MemTool, a short-term memory framework enabling LLM agents to dynamically manage tools or MCP server contexts across multi-turn conversations. MemTool offers three agentic architectures: 1) Autonomous Agent Mode, granting full tool management autonomy, 2) Workflow Mode, providing deterministic control without autonomy, and 3) Hybrid Mode, combining autonomous and deterministic control. Evaluating each MemTool mode across 13+ LLMs on the ScaleMCP benchmark, we conducted experiments over 100 consecutive user interactions, measuring tool removal ratios (short-term memory efficiency) and task completion accuracy. In Autonomous Agent Mode, reasoning LLMs achieve high tool-removal efficiency (90-94% over a 3-window average), while medium-sized models exhibit significantly lower efficiency (0-60%). Workflow and Hybrid modes consistently manage tool removal effectively, whereas Autonomous and Hybrid modes excel at task completion. We present trade-offs and recommendations for each MemTool mode based on task accuracy, agency, and model capabilities.

Autonomous AI agents tackling Long Horizon Tasks depend on marketplace skills that are certified one at a time: a scanner returns a safety verdict for each skill and declares the ecosystem safe if every package passes. We show that this assumption fails under skill composition. A skill may pass the per-skill scanner individually yet participate in a risky composition when an agent connects its outputs, capabilities, or side effects with those of other scanner-passing skills. This makes skill composition risk a path level property rather than a node level property, explaining why existing skill scanners that inspect individual packages achieve limited interception. To study this threat, we present CompoSkill, a framework that constructs skill composition attacks through a dual attacker system. The white-box attacker knows the victim's installed skill pool and directly injects explicit skill-id sequences; the black-box attacker knows only a role profile, downloads the top marketplace skills for that scenario, builds a Skill Composition Graph, and searches for high risk chains whose implicit lures never name skill identifiers. We further construct CompoSkill-Bench, a benchmark of 1,140 records built from long-horizon professional workflows across five threats and six scenarios on OpenClaw and Nanobot. CompoSkill achieves risk Chain Formation Rates (CFR) up to 83.3% in the white box setting and 80.6% in the black box setting, while existing skill scanners block only a limited fraction of the risky compositions. Finally, we observe a bridge-bonus-then-hop-decay pattern: a bridge skill can increase attack success, but Attack Success Rate (ASR) decreases once additional hops make the risk chain longer than three skills. These results expose a systematic gap in single skill certification for autonomous AI agents.

The integration of Large Language Models (LLMs) into software engineering has driven a transition from traditional rule-based systems to autonomous agentic systems capable of solving complex problems. However, systematic progress is hindered by a lack of comprehensive understanding of how benchmarks and solutions interconnect. This survey addresses this gap by providing the first holistic analysis of LLM-powered software engineering, offering insights into evaluation methodologies and solution paradigms. We review over 150 recent papers and propose a taxonomy along two key dimensions: (1) Solutions, categorized into prompt-based, fine-tuning-based, and agent-based paradigms, and (2) Benchmarks, including tasks such as code generation, translation, and repair. Our analysis highlights the evolution from simple prompt engineering to sophisticated agentic systems incorporating capabilities like planning, reasoning, memory mechanisms, and tool augmentation. To contextualize this progress, we present a unified pipeline illustrating the workflow from task specification to deliverables, detailing how different solution paradigms address various complexity levels. Unlike prior surveys that focus narrowly on specific aspects, this work connects 50+ benchmarks to their corresponding solution strategies, enabling researchers to identify optimal approaches for diverse evaluation criteria. We also identify critical research gaps and propose future directions, including multi-agent collaboration, self-evolving systems, and formal verification integration. This survey serves as a foundational guide for advancing LLM-driven software engineering. We maintain a GitHub repository that continuously updates the reviewed and related papers at https://github.com/lisaGuojl/LLM-Agent-SE-Survey.

Interactive large language model agents have advanced rapidly, but most remain specialized to a single environment and fail to adapt robustly to other environments. Model merging offers a training-free alternative by integrating multiple experts into a single model. In this paper, we propose Agent-Role Merging (ARM), an activation-guided, role-conditioned neuron transplantation method for model merging in LLM agents. ARM improves existing merging methods from static natural language tasks to multi-turn agent scenarios, and over the generalization ability across various interactive environments. This is achieved with a well designed 3-step framework: 1) constructing merged backbones, 2) selection based on its role-conditioned activation analysis, and 3) neuron transplantation for fine-grained refinements. Without gradient-based optimization, ARM improves cross-benchmark generalization while enjoying efficiency. Across diverse domains, the model obtained via ARM merging outperforms prior model merging methods and domain-specific expert models, while demonstrating strong out-of-domain generalization.

Skill-distillation pipelines learn reusable rules from LLM agent trajectories, but they lack a key signal: how much each step costs. Without per-step cost, a pipeline cannot distinguish adding a missing step to fix a bug from removing an expensive step that never affected the outcome. We introduce ClawTrace, an agent tracing platform that records every LLM call, tool use, and sub-agent spawn during an agent session and compiles each session into a TraceCard: a compact YAML summary with per-step USD cost, token counts, and redundancy flags. Built on ClawTrace, CostCraft is a distillation pipeline that reads TraceCards and produces three types of skill patches. Preserve patches keep behaviors that led to success. Prune patches remove expensive steps that did not matter, each backed by a counterfactual argument against a named high-cost step. Repair patches fix failures grounded in oracle evidence. Ablations on 30 held-out SpreadsheetBench tasks show that both cost attribution and prune patches independently reduce quality regressions. When the same skill is applied to 30 unrelated SkillsBench tasks, an unexpected asymmetry emerges: prune rules transferred across benchmarks and cut median cost by 32%, while preserve rules, trained on benchmark-specific conventions, caused regressions on new task types. We release ClawTrace and TraceCards as open infrastructure for cost-aware agent research.

The capability of LLM agents to function as the ``brain'' of a system fundamentally expands the scope of analysis beyond a standalone model. Consequently, safety is no longer only about input--output content alignment. It also concerns system behavior and real-world execution outcomes. However, the current literature is fragmented across attack types, applications, and benchmarks. This makes it hard to explain why failures such as prompt injection, tool misuse, and memory poisoning often share the same structural cause, and how they spread through an agent workflow. In this survey, we treat isolation as a first-class principle for LLM-agent system safety. By isolation, we refer to the separation of user inputs, tool access, execution channels, inter-agent communication, and environment-originated context. We organize the literature with a boundary-centric taxonomy of five boundaries: user-agent, agent-tool, agent-execution, agent-agent, and system-environment. This view helps identify where the loss of isolation first occurs, how compromise propagates across boundaries, and which defenses are most relevant at each interface. We also summarize cross-boundary failure paths, discuss open challenges, and outline a research agenda for isolation-by-construction in future agent systems.

A prevalent assumption in LLM agent deployment holds that more structured harnesses universally improve reliability, and that higher-capability models need proportionally less structural guidance -- together implying a monotone inverse relationship between model capability tier and optimal harness complexity. We test this hypothesis through a controlled 432-run experiment crossing six models across four capability tiers with three harness conditions (light, balanced, strict) on HEAT-24, a 24-task synthetic benchmark with git-based workspace verification. Our results refute the monotone inverse relationship on two fronts. First, for the frontier chat model evaluated (Gemini 2.5 Flash), increased harness verbosity lowers VTSR by 29-38 percentage points -- a harness-complexity paradox. Second, for the frontier reasoning model evaluated (Qwen3.5-122B, extended thinking enabled), strict harness achieves the highest VTSR (91.7%) and the lowest latency, the opposite of the prediction. Within the constrained tier, a 2B model (Gemma4:e2B) matches strong-open-tier stability at 91.7% across all harnesses. Because each tier is represented by a single model in this study, these results should be interpreted as model-specific observations; harness sensitivity appears non-monotone across the models evaluated, and depends critically on model type (chat vs. reasoning). We introduce a six-label failure taxonomy showing that format_violation dominates capable-model failures while wrong_file dominates low-capability failures, and we derive practical tier-aware harness selection guidelines.

Answering complex natural language questions often necessitates multi-step reasoning and integrating external information. Several systems have combined knowledge retrieval with a large language model (LLM) to answer such questions. These systems, however, suffer from various failure cases, and we cannot directly train them end-to-end to fix such failures, as interaction with external knowledge is non-differentiable. To address these deficiencies, we define a ReAct-style LLM agent with the ability to reason and act upon external knowledge. We further refine the agent through a ReST-like method that iteratively trains on previous trajectories, employing growing-batch reinforcement learning with AI feedback for continuous self-improvement and self-distillation. Starting from a prompted large model and after just two iterations of the algorithm, we can produce a fine-tuned small model that achieves comparable performance on challenging compositional question-answering benchmarks with two orders of magnitude fewer parameters.

Large language models are increasingly deployed as agents, but reliable agentic behavior requires more than next-token prediction. At inference time, it is preferred that an agent can decide whether to proceed with its current reasoning, defer to a stronger model, request additional information, invoke external tools, or abstain under the given setup. Existing approaches address these decisions through prompt-level routing, external orchestration, or task-specific fine-tuning, which primarily rely on input-side signals, and are often costly and difficult to maintain as model backbones evolve. We ask whether such control decisions can be inferred directly from a model's latent generation process. We introduce Multi-Head Latent Control, a lightweight layer that reads hidden-state trajectories from a frozen LLM or VLM to produce deployment-time control signals. A Capability Head predicts whether the current model can solve the instance or should defer to a stronger collaborator, while a Resolution Head predicts appropriate resolution decision Clarification, Tool Use, Abstention, or Direct Answering. Both heads are trained only on latent traces from the same frozen LLM backbone, enabling post hoc adaptation without modifying the model. Across language and vision-language settings, Multi-Head Latent Control consistently improves the quality-cost tradeoff of multi-model systems, enabling early handoff from partial generations and more accurate intervention decisions. In routed execution (small + large model), it reduces large-model usage by up to 90.7 percent on AndroidWorld and 27-53 percent on average across benchmarks, while retaining most of large-model performance. Additionally, the learned control signals improve tool-use decision quality, yielding up to +158 percent relative score gain and 65.5 percent fewer missed-required tool calls.

Language model agents now execute bounded tasks reliably. Whether they can sustain effective decision-making over long horizons, where actions have cumulative consequences and the environment responds to their choices, remains largely unmeasured. FM-Bench (Football Management Benchmark) measures this. An LLM agent runs a football club for 20 in-game years through 26 tools and roughly 340 to 400 decision stops. It drafts a squad on the same budget as every rival, trades players, negotiates contracts, invests in facilities and youth, sets lineups, and answers to a board that can fire it, while a deterministic engine accumulates every year into one final score with no LLM judge or human rater. The solo track plays each of 15 frontier models against a frozen scripted world, and the Arena places the same models plus a scripted anchor in one shared 20-year world; to our knowledge, the first head-to-head evaluation at this scale. We measure six behavioral capabilities behind the score. Across three seeds, all 15 models complete every horizon while the blind scripted baselines die out in most of theirs, and claude-fable-5 tops the solo board on mean score and the Arena, where the title nonetheless rotates among ten models. Neither scale, price, nor vendor predicts the order; the order settles only late in the horizon, and the best first-play human lands only at the bottom of the model board. What separates the models is managerial behavior rather than computation. Higher-scoring models reduce slow-payoff investment near the end, keep cash invested rather than idle, and open renewals well before the deadline, while token spend predicts nothing. No model learns the market's hidden prices from hundreds of rejected bids, and self-managed memory fails in two opposite modes: an archive that only grows or a plan rewritten every season. Code is available at https://github.com/Analogy-AI/fm-bench.

Personal agents maintain memories, learned skills, tool configurations, and policy state that evolve with each user. Existing agent benchmarks often evaluate these capabilities in isolation: tool benchmarks test invocation under fixed APIs, memory benchmarks test recall or forgetting, and safety benchmarks test static policy compliance. We argue that personal-agent evaluation requires a different protocol: replaying the same temporal intervention across different persistent user-conditioned states and measuring how failures propagate across agent components. We formalize this requirement as four conditions: explicit temporal intervention, persistent state across the intervention, induced cross-dimensional effects, and variation in user-conditioned state. A focused audit of public benchmark protocols selected by explicit inclusion criteria identifies several close cases. Under our explicitly narrow operationalization, we did not find a protocol in that audited set satisfying all four conditions. This claim is scoped as a focused gap analysis with bounded literature coverage. This position paper proposes a minimal benchmark design and candidate reporting metrics for user-conditioned adaptation. The result is a concrete design requirement for future personal-agent evaluation, with metrics used as reporting tools for that requirement.

We introduce PhysicianBench, a benchmark for evaluating LLM agents on physician tasks grounded in real clinical setting within electronic health record (EHR) environments. Existing medical agent benchmarks primarily focus on static knowledge recall, single-step atomic actions, or action intent without verifiable execution against the environment. As a result, they fail to capture the long-horizon, composite workflows that characterize real clinical systems. PhysicianBench comprises 100 long-horizon tasks adapted from real consultation cases between primary care and subspecialty physicians, with each task independently reviewed by a separate panel of physicians. Tasks are instantiated in an EHR environment with real patient records and accessed through the same standard APIs used by commercial EHR vendors. Tasks span 21 specialties (e.g., cardiology, endocrinology, oncology, psychiatry) and diverse workflow types (e.g., diagnosis interpretation, medication prescribing, treatment planning), requiring an average of 27 tool calls per task. Solving each task requires retrieving data across encounters, reasoning over heterogeneous clinical information, executing consequential clinical actions, and producing clinical documentation. Each task is decomposed into structured checkpoints (670 in total across the benchmark) capturing distinct stages of completion graded by task-specific scripts with execution-grounded verification. Across 13 proprietary and open-source LLM agents, the best-performing model achieves only 46% success rate (pass@1), while open-source models reach at most 19%, revealing a substantial gap between current agent capabilities and the demands of real-world clinical workflows. PhysicianBench provides a realistic and execution-grounded benchmark for measuring progress toward autonomous clinical agents.

Skill-based prompting has become a practical mechanism for improving large language model (LLM) agents, yet existing skill acquisition methods often treat skills as experience summaries, memory entries, or direct summaries of successful demonstrations. This creates a mismatch for weaker student agents: when a student fails because it lacks task knowledge or operational strategy, its failed trajectory may not contain enough evidence to infer the missing behavior, while the teacher trajectory may be too implicit to be internalized as reusable guidance. We propose SKILL-KD, a contrastive skill distillation framework that treats skills as an explicit distillation medium between agents of different capabilities. Given a student failure and the teacher trajectory on the same task, SKILL-KD distills their actionable discrepancy into a textual skill patch, evaluates the patch by re-running the student, and iteratively refines the patch when the student still fails. To prevent repeated local updates from causing skill drift, SKILL-KD further maintains trace-linked edit histories and performs Drift-Aware Skill Consolidation, deciding whether each patch should add a new rule, delete or modify an existing rule, or be skipped. Across five agent benchmarks and two student settings, SKILL-KD consistently improves frozen student agents over fixed-model adaptation baselines.

Agent benchmarks are growing fast, but no single benchmark touches more than four or five of the dimensions that deployment exposes. This paper aggregates the largest coordinated deep-dive of one MCP-based industrial-agent benchmark to date: fourteen parallel implementation studies covering new asset classes (including a multi-modal visual extension), alternative orchestrations, retrieval strategies, reasoning modes, infrastructure optimizations, and evaluation-methodology probes. Consolidating those studies with seven prior agent benchmarks, we argue that aggregate-score leaderboards systematically underspecify deployed-agent evaluation. Rankings derived from aggregate scores do not transfer to out-of-distribution settings; recent public-to-hidden competition retrospectives provide direct empirical evidence of this rank instability. We propose ranking configurations by predictive validity, the correlation between in-sample and out-of-sample rank, rather than in-sample mean, and report a twelve-tier measurement apparatus that exposes the deployment-relevant dimensions HELM and its agent-era successors collapse. The position is operationalized through three falsifiable out-of-distribution criteria with explicit thresholds; existing evidence partly supports it but is too thin to confirm. We close with a pre-registered pilot design and a field-level vision for what the next generation of agentic benchmarks should report.

Financial agents are expected to answer factual investment queries through open search, but financial multi-hop retrieval differs from general multi-hop QA in a crucial way: each intermediate answer is an exact typed constraint, such as an entity, date, event, or numerical indicator, that determines the next query. This strict answer-conditioned dependency makes small intermediate errors cascade into plausible but wrong search trajectories. We introduce FinHopper, a benchmark for complex financial agents with controlled reasoning depth from 1 to 6 hops, covering stocks, funds, and macroeconomic facts, with gold intermediate answers for process-level evaluation. To construct FinHopper at scale, we design a state-transition-based synthesis framework that composes verified atomic financial QA patterns into dependency-preserving chains, followed by targeted in-synthesis verification, model-ensemble checking, and public-search validation by financial experts. Experiments with 20 LLM-agent configurations show that public search tools are necessary but insufficient: performance drops sharply as hop depth increases, the best model remains far below human experts, and process-aware scoring reveals partial reasoning progress hidden by final-answer-only evaluation. The benchmark is available at https://github.com/justajustin/FinHopper/.

Reinforcement learning (RL) holds significant promise for training LLM agents to handle complex, goal-oriented tasks that require multi-step interactions with external environments. However, a critical challenge when applying RL to these agentic tasks arises from delayed rewards: feedback signals are typically available only after the entire task is completed. This makes it non-trivial to assign delayed rewards to earlier actions, providing insufficient guidance regarding environmental constraints and hindering agent training. In this work, we draw on the insight that the ultimate completion of a task emerges from the cumulative progress an agent makes across individual steps. We propose Stepwise Progress Attribution (SPA), a general reward redistribution framework that decomposes the final reward into stepwise contributions, each reflecting its incremental progress toward overall task completion. To achieve this, we train a progress estimator that accumulates stepwise contributions over a trajectory to match the task completion. During policy optimization, we combine the estimated per-step contribution with a grounding signal for actions executed in the environment as the fine-grained, intermediate reward for effective agent training. Extensive experiments on common agent benchmarks (including Webshop, ALFWorld, and VirtualHome) demonstrate that SPA consistently outperforms the state-of-the-art method in both success rate (+2.5\% on average) and grounding accuracy (+1.9\% on average). Further analyses demonstrate that our method remarkably provides more effective intermediate rewards for RL training. Our code is available at https://github.com/WangHanLinHenry/SPA-RL-Agent.

One paradigm of language model (LM) fine-tuning relies on creating large training datasets, under the assumption that high quantity and diversity will enable models to generalize to novel tasks after post-training. In practice, gathering large sets of data is inefficient, and training on them is prohibitively expensive; worse, there is no guarantee that the resulting model will handle complex scenarios or generalize better. Moreover, existing techniques rarely assess whether a training sample provides novel information or is redundant with the knowledge already acquired by the model, resulting in unnecessary costs. In this work, we explore a new test-time self-improvement method to create more effective and generalizable agentic LMs on-the-fly. The proposed algorithm can be summarized in three steps: (i) first it identifies the samples that model struggles with (self-awareness), (ii) then generates similar examples from detected uncertain samples (self-data augmentation), and (iii) uses these newly generated samples at test-time fine-tuning (self-improvement). We study two variants of this approach: Test-Time Self-Improvement (TT-SI), where the same model generates additional training examples from its own uncertain cases and then learns from them, and contrast this approach with Test-Time Distillation (TT-D), where a stronger model generates similar examples for uncertain cases, enabling student to adapt using distilled supervision. Empirical evaluations across different agent benchmarks demonstrate that TT-SI improves the performance with +5.48% absolute accuracy gain on average across all benchmarks and surpasses other standard learning methods, yet using 68x less training samples. Our findings highlight the promise of TT-SI, demonstrating the potential of self-improvement algorithms at test-time as a new paradigm for building more capable agents toward self-evolution.

Existing large language models (LLMs) evaluations use fixed-difficulty benchmarks that cannot adapt as models improve, and rarely isolate specific cognitive processes. We introduce Working Memory Fidelity-Active Manipulation (WMF-AM), a probe of cumulative state tracking, the ability to maintain and update intermediate results across K sequential operations within a single query, without a scratchpad. Unlike multi-step agent benchmarks that stress task orchestration, WMF-AM isolates within-pass cumulative load by parameterizing depth K. The core probe uses arithmetic accumulation on 28 models from 12 families (0.5B to frontier); a matched non-arithmetic extension (permissions, schedules, inventories) confirms the design generalizes beyond arithmetic. Three construct-isolation ablations confirm that cumulative load, not arithmetic skill or entity tracking, drives difficulty. We release WMF-AM as a lightweight, recalibratable diagnostic for characterizing where models degrade under cumulative load. Code and data can be accessed at https://github.com/dengzhe-hou/WMF-AM

LLM agents act in external environments where each action changes the state that later decisions condition on, and where a single wrong step can waste interaction budget or trigger irreversible side effects long before the final failure is observed. Reliable deployment therefore requires step-level confidence estimation: a calibrated probability that each proposed action is productive, available before the action is executed. Existing LLM confidence estimators are designed to score a response from the given prompt, but agent confidence also depends on execution consequences: whether similar actions in similar situations actually advanced the task after the environment responded. We introduce the \method (\methodshort), a self-evolving critic framework in which an LLM critic accumulates evidence from its own past judgments and their observed consequences. After each trajectory, a hindsight LLM that sees the full execution feedback votes on whether each step was productive. The resulting pseudo-labels populate a memory bank from which related productive and unproductive experiences are retrieved into the critic's prompt whenever a similar step recurs. \methodshort requires no training and uses no ground truth step labels. Across three agent benchmarks and three critic backbones, \methodshort attains the best calibration (ECE and Brier) and ranking (AUC) in every dataset--critic combination, reducing ECE by up to 54% relative to the strongest training-free baseline.

Deception is a fundamental challenge for multi-agent reasoning: effective systems must strategically conceal information while detecting misleading behavior in others. Yet most evaluations reduce deception to static classification, ignoring the interactive, adversarial, and longitudinal nature of real deceptive dynamics. Large language models (LLMs) can deceive convincingly but remain weak at detecting deception in peers. We present WOLF, a multi-agent social deduction benchmark based on Werewolf that enables separable measurement of deception production and detection. WOLF embeds role-grounded agents (Villager, Werewolf, Seer, Doctor) in a programmable LangGraph state machine with strict night-day cycles, debate turns, and majority voting. Every statement is a distinct analysis unit, with self-assessed honesty from speakers and peer-rated deceptiveness from others. Deception is categorized via a standardized taxonomy (omission, distortion, fabrication, misdirection), while suspicion scores are longitudinally smoothed to capture both immediate judgments and evolving trust dynamics. Structured logs preserve prompts, outputs, and state transitions for full reproducibility. Across 7,320 statements and 100 runs, Werewolves produce deceptive statements in 31% of turns, while peer detection achieves 71-73% precision with ~52% overall accuracy. Precision is higher for identifying Werewolves, though false positives occur against Villagers. Suspicion toward Werewolves rises from ~52% to over 60% across rounds, while suspicion toward Villagers and the Doctor stabilizes near 44-46%. This divergence shows that extended interaction improves recall against liars without compounding errors against truthful roles. WOLF moves deception evaluation beyond static datasets, offering a dynamic, controlled testbed for measuring deceptive and detective capacity in adversarial multi-agent interaction.

Modern analyst agents must reason over complex, high token inputs, including dozens of retrieved documents, tool outputs, and time sensitive data. While prior work has produced tool calling benchmarks and examined factuality in knowledge augmented systems, relatively little work studies their intersection: settings where LLMs must integrate large volumes of dynamic, structured and unstructured multi tool outputs. We investigate LLM failure modes in this regime using crypto as a representative high data density domain. We introduce (1) CryptoAnalystBench, an analyst aligned benchmark of 198 production crypto and DeFi queries spanning 11 categories; (2) an agentic harness equipped with relevant crypto and DeFi tools to generate responses across multiple frontier LLMs; and (3) an evaluation pipeline with citation verification and an LLM as a judge rubric spanning four user defined success dimensions: relevance, temporal relevance, depth, and data consistency. Using human annotation, we develop a taxonomy of seven higher order error types that are not reliably captured by factuality checks or LLM based quality scoring. We find that these failures persist even in state of the art systems and can compromise high stakes decisions. Based on this taxonomy, we refine the judge rubric to better capture these errors. While the judge does not align with human annotators on precise scoring across rubric iterations, it reliably identifies critical failure modes, enabling scalable feedback for developers and researchers studying analyst style agents. We release CryptoAnalystBench with annotated queries, the evaluation pipeline, judge rubrics, and the error taxonomy, and outline mitigation strategies and open challenges in evaluating long form, multi tool augmented systems.

Evaluating LLM agents in realistic service scenarios requires complex task dependencies, imperfect user behavior, and an evaluation that accommodates multiple valid solutions. We introduce CRAB-Bench (Constraint-based Realistic Agent Benchmark) and RUSE (Realistic User Simulation Engine) to address this gap. CRAB-Bench generates tasks via a constraint graph over multiple interdependent entities with structured distractors, requiring agents to reason carefully over thousands of misleading candidates where only a tiny fraction of solutions are valid. RUSE replaces cooperative, template-like simulators with realistic users grounded in human behavioral studies, instantiated across diverse personas and four behavioral dimensions. Experiments on four frontier LLM agents show that the best model achieves only 61% pass@1 on CRAB-Bench, and switching to RUSE causes further drops of up to 57%, concentrated in task-solving ability rather than conversational quality. Information Disclosure is the most damaging behavioral dimension, and agents interacting with RUSE are less likely to admit mistakes, instead masking errors through implicit corrections.