Agent 进展

Large language model (LLM)-based machine translation has advanced cross-cultural communication, yet it still struggles with culture-loaded words (CLWs) in ancient Chinese texts. The challenge extends beyond lexical alignment to deciding when and how culture-dependent knowledge should be explicated for readers lacking relevant background. Literal translation often preserves surface forms while missing underlying concepts, whereas over-explicitation harms conciseness and readability. To address this problem, we formulate CLW translation as a selective explicitation task and propose MACAT, a Multi-Agent Culture-Aware Translation framework that dynamically identifies culturally salient phrases and injects concise explanatory knowledge when necessary. MACAT further incorporates a quality-aware reranking module for candidate selection and a multi-round evaluation agent that assesses translations across terminological precision, readability, fidelity, cultural preservation, and cultural explicitation. Experiments on traditional Chinese medicine (TCM) classics and the Analects show that, under a unified GPT-5.4 evaluation setting, MACAT consistently outperforms both the backbone model and general-purpose MT baselines on 100 TCM documents and a 20-chapter subset of the Analects.

Drug-information question answering is a high-stakes setting where hallucinated facts can mislead clinical decision-making and the provenance of each cited fact matters as much as the fact itself. We present DrugClaw, a multi-agent retrieval-augmented system that queries a registry of drug and pharmacovigilance skills via a reflection-driven state-machine workflow and returns answers grounded in primary regulatory or peer-reviewed records. We also contribute DrugAudit, a 3,772-item authority-aware benchmark with an evaluation panel that scores upstream-of-gold source match, token-level semantic snippet overlap, and citation faithfulness under a dual-judge LLM-as-judge protocol with inter-judge kappa = 0.88 (almost-perfect). Across DrugAudit plus drug-related subsets of MedQA (751) and PubMedQA (512), DrugClaw is top-1 on every column of the headline table: composite Evidence Index under both judges, judge-mediated answer correctness, primary-source rate (0.918, +10.1 pp over next-best), faithfulness (0.887, +5.9 pp), MedQA (0.920), and PubMedQA (0.693).

Reusable skills are a key mechanism for extending agent capabilities, allowing agents to accumulate experience and solve increasingly complex tasks. Yet most existing skill-learning methods store reusable experience as text-only assets, such as instructions, reasoning traces, or summarized trajectories. We argue that this text-only paradigm creates a fundamental bottleneck for visual-centric tasks, where reusable knowledge often depends on spatial layout, visual grounding, fine-grained appearance, and localized state changes. To address this limitation, we propose \NAME, a multimodal skill paradigm that combines declarative textual logic with explicit visual support. We distinguish three reusable forms: static priors for stable spatial conventions, dynamic priors for in-situ visual working memory, and interleaved visual skills that bind ordered text steps to the source frames, screenshots, or page regions that justify them. Rather than only describing what to do, visual skills also encode where to look, how to inspect, and how to verify visual outcomes. To scale visual-skill construction, we introduce \SYSTEM, an automatic system that converts agent experience into reusable multimodal skills by preserving textual reasoning, spatial references, visual boundaries, and interaction patterns from task trajectories. Experiments on GUI and other visual-centric tasks show that visual skills consistently outperform text-only skills, particularly when success requires spatial correspondence, visual evidence, and state-aware interaction. These results support our central position: reusable agent skills should go beyond text and become multimodal assets for future multimodal agents.

Computer-use agents extend language models from text generation to sustained interaction with files, terminals, browsers, and external tools. This shift creates safety risks that are difficult to detect from isolated prompts or final responses, because harm often emerges only through multi-step execution traces whose individual actions appear locally benign. We introduce BraveGuard, a self-evolving defense framework for training guard models from open-world threat signals and realistic agent trajectories. BraveGuard mines recent research sources to identify emerging risks and attack patterns, instantiates them as executable computer-use tasks, collects agent rollouts, and derives trajectory-level supervision for guard model training. As new threats and validation failures appear, the pipeline can be repeated, yielding an adaptive defense loop rather than a static, benchmark-driven training process. We instantiate BraveGuard by training multiple guard backbones, including Qwen3-Guard and Llama-Guard variants, and evaluate the resulting guards on trajectory-level agent-safety benchmarks. BraveGuard consistently improves safety detection across computer-use trajectories. On AgentHazard, it substantially improves detection accuracy over off-the-shelf guard models, with accuracy increasing from 38.79% to 82.38% under the averaged guard-model setting. These results show that guard supervision grounded in open-world threat discovery and realistic agent execution can improve safety monitoring beyond fixed taxonomies and synthetic prompt-level data. BraveGuard offers a scalable path toward adaptive defenses for computer-use agents facing evolving real-world risks.

Recent text-clustering methods use large language models to propose a cluster taxonomy from a corpus and then assign each text to it. These pipelines are fundamentally programmatic: the sequence of LLM calls and the rules for stopping, merging, and splitting clusters are fixed in code in advance, so they generalise poorly across corpora of different structure and cannot easily incorporate user-supplied constraints such as a target cluster count or a clustering intent. We propose an agentic alternative in which an orchestrator LLM inspects the state of the discovery process at each step and dispatches one of a small set of specialised agents - proposer, synthesizer, auditor, investigator, and critic - adapting the pipeline to the corpus rather than executing a fixed one. On seven public text-clustering benchmarks the method achieves state-of-the-art performance, beating the strongest prior LLM baseline by up to 32% in ARI.

Large language model (LLM) agents increasingly rely on reusable external skills to solve long-horizon interactive tasks. Existing training-free skill adaptation pipelines usually update skills from full trajectories or session-level feedback, which makes failure attribution coarse and often produces unstable or overly broad revisions. We propose SkillAdaptor, a training-free step-level skill adaptation framework with explicit failure attribution, and it can plug into OpenClaw-class agent harnesses. Given a failed trajectory, SkillAdaptor identifies a first actionable fault step, links responsibility to candidate skills, and applies targeted updates under explicit acceptance checks while keeping the backbone frozen. We evaluate on WebShop, PinchBench, and Claw-Eval with Kimi-K2.5, GLM-5, and GPT-5.2. SkillAdaptor improves over no-skill and skill-adaptation baselines on all three suites, with the largest single-metric improvements of +1.5 points on PinchBench Avg Score%, +1.8 on Claw-Eval Avg Score, and +1.7 on WebShop success rate. These results indicate that step-level attribution supports more stable and auditable training-free skill maintenance\footnote{The code will be released at https://github.com/zjunlp/SkillAdaptor.}.

Tool-using multi-agent large language model (LLM) systems spend computation through model tokens, tool calls, retries, and code execution before producing an answer. When a run fails, final-answer evaluation reveals the endpoint but usually not the point at which the trajectory stopped making recoverable progress. This paper introduces a failure-aware observability framework for diagnosing wasted computation in multi-agent LLM traces. The framework maps recurring failure modes to online trace signals, including tool reliability, execution recovery, orchestration loops, evidence availability, information change, and budget pressure. We instantiate the framework in a three- agent question-answering system and evaluate it on 165 GAIA validation traces under identical execution caps. Operational failures remain common: 22/53 level-1 runs, 33/86 level-2 runs, and 12/26 level-3 runs fail to produce a usable final answer. The traces expose different mechanisms behind these outcomes, including insufficient evidence, repeated-action loops, max-step termination, tool-failure streaks, and execution calls that succeed without useful output. Mean token use rises from 8,152 tokens at level 1 to 16,389 tokens at level 3, while evidence availability and sentence-level support diverge. A cached 10-trace LLM-judge grounding audit shows that cheap online signals and deeper semantic metrics capture complementary layers of failure. The results position failure-aware observability as a diagnostic layer between raw execution logs and final-answer accuracy.

Software architecture design is a critical yet inherently complex and knowledge-intensive phase that requires balancing competing quality attributes and adapting to evolving requirements. Traditionally, this process has been time-consuming, labor-intensive, and heavily reliant on architects, often resulting in limited exploration of alternative architectural decompositions and styles, especially under the pressures of agile development. While LLM-based agents have shown promising performance across various software engineering tasks, their application to architecture design remains relatively scarce and requires systematic exploration. To address these challenges, we proposed MAAD (Multi-Agent Architecture Design), a knowledge-driven framework that orchestrates four specialized agents (i.e., Analyst, Modeler, Designer and Evaluator) to autonomously and collaboratively transform requirements specifications into comprehensive, multi-view architectural blueprints with quality attribute assessments. MAAD incorporates RAG to inject recognized architectural standards and patterns into the workflow and leverages a hierarchical memory mechanism that captures design history for iterative refinement. We evaluated MAAD through comparative experiments against MetaGPT, using quantitative architecture-level metrics across 10 case studies and qualitative feedback from industry architects on 10 real-world specifications. Results show that MAAD generates more complete, modular, and traceable architectures than the baseline, and its dedicated Evaluator agent autonomously produces structured quality evaluation reports that significantly reduce manual validation efforts. Furthermore, we found that the quality of the generated architecture heavily depends on the underlying LLM's reasoning capacity, with GPT-5.2 and Qwen3.5 outperforming other models across most evaluation settings.

Experience learning has achieved promising results in enhancing LLM agent planning and reasoning by integrating past interactions as reusable knowledge. However, existing methods remain confined to explicit text space, retrieving experiences via semantic similarity and concatenating them into the context window, leading to substantial token overhead and a decoupled architecture that separates retrieval from generation. To address these limitations, we propose ExpWeaver, a framework that enables LLM agents to learn from experience via latent retrieval-augmented generation, without requiring a separate RAG module. ExpWeaver encodes experiences using the LLM's own hidden states, retrieves relevant experiences directly in latent space at each decoding step, and integrates them through cross-attention aggregation and gated residual mechanisms. The entire pipeline is optimized end-to-end with reinforcement learning, supporting both generative and ranking tasks. We evaluate ExpWeaver on 13 diverse tasks spanning question answering, reasoning, coding, scientific prediction, and recommendation. Results demonstrate that ExpWeaver achieves state-of-the-art performance on 12 out of 13 tasks, outperforming the strongest baseline by over 6.8%; maintains token efficiency comparable to non-retrieval baselines while text-based retrieval methods require 1.5 to 2 times more tokens; and exhibits superior cross-domain generalization, outperforming the strongest baseline by 16.32% under zero-shot transfer and 15.21% under few-shot transfer. Our code for ExpWeaver is released at https://github.com/ulab-uiuc/ExpWeaver.

Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for enhancing the reasoning capabilities of large language models (LLMs). However, its effectiveness is substantially hindered by the prevalence of ineffective training data: many sampled prompts yield response groups that are either entirely correct or entirely incorrect, resulting in zero-variance rewards and limited learning signals. Recent state-of-the-art methods address this issue through extensive LLM rollouts to filter ineffective samples, but at the cost of considerable computational overhead. Alternative approaches, including predictive sampling and trajectory replay, aim to improve data efficiency but often remain insufficient and may introduce additional issues such as systematic bias or suboptimal constraints. To address these limitations, we propose Group Prioritized Off-Policy Optimization (POPO), a simple yet effective framework that fully exploits effective training batches without additional rollout overhead. POPO comprises two key components: prioritized group replay and decoupled off-policy optimization. The former replaces ineffective on-policy groups with effective off-policy groups via a recency-based replay mechanism that jointly considers sample quality and the degree of off-policiness. To further mitigate the off-policy gap, POPO employs decoupled importance sampling to correct off-policy bias while maintaining stable policy updates under consistent trust-region constraints. Empirical evaluations across diverse reasoning tasks, including mathematics, planning, and visual geometry, demonstrate that POPO substantially accelerates RL finetuning and achieves strong reasoning performance with significantly fewer rollouts.

Mastering terminal environments requires language agents capable of multi-step planning, feedback-grounded execution, and dynamic state adaptation. However, training such agents is currently bottlenecked by a reliance on scraped external repositories, which limits domain diversity, environment controllability, and the targeting of specific capability deficits. We introduce LiteCoder-Terminal-Gen, a zero-dependency synthesis pipeline that autonomously generates executable and verifiable terminal training environments directly from domain specifications. Using this framework, we construct two large-scale resources: LiteCoder-Terminal-SFT, comprising 11,255 expert trajectories across 10 domains, and LiteCoder-Terminal-RL, featuring 602 verifiable environments for trajectory-level preference optimization. Supervised fine-tuning of Qwen-family models on our SFT dataset yields agents that significantly outperform their base counterparts. Notably, our 32B variant achieves 29.06%, 18.54%, and 34.00% pass@1 on Terminal Bench 1.0, 2.0, and Pro, respectively. Furthermore, applying Direct Multi-turn Preference Optimization (DMPO) on our RL environments yields additional performance gains. These results systematically demonstrate that fully synthetic, executable environments offer a scalable and verifiable supervision signal for mastering complex, real-world command-line workflows.

Given a strategically complex board game, human players can quickly learn to devise strategies after playing a few rounds. Autonomous agents require similar capabilities in realistic interactive environments, yet existing agent benchmarks often fail to fully capture such strategic and evolving decision-making scenarios. We present PTCG-Bench, a benchmark built on the Pok'{e}mon Trading Card Game (PTCG) that evaluates LLM agents at two complementary levels: (1) their decision-making performance within a single complex environment, and (2) their ability to self-evolving through accumulated experience. We further include a modular harness ablation to better interpret agent performance without conflating it with model capability. Our experiments show that, although LLM agents can achieve non-trivial gameplay performance, sustained and stable self-evolution remains challenging, and performance is sensitive to harness design. We hope that PTCG-Bench will facilitate future research on harness-aware and self-evolving agents in realistic interactive environments.

LLM agents acting in structured environments fail in operational rather than conversational ways, and reliability depends on procedural knowledge of the environment. Prior self-improvement methods accumulate natural-language guidance without checking that each new item preserves previously correct behavior, so a note that fixes one trajectory can silently regress another. We introduce GRASP (Gated Regression-Aware Skill Proposer), which treats agent improvement as a sequence of edits to a bounded skill library, admitting each candidate only if it produces a net improvement on a balanced held-out probe under a hard regression budget. We evaluate GRASP across five base models (gpt-oss-120b, DeepSeek V4 Flash, Gemini 3.1 Flash Lite, GPT-4.1, GPT-5.4) on two FHIR-based clinical benchmarks. On MedAgentBench, GRASP lifts gpt-oss-120b from 40.6% to 88.8%, exceeds the strongest of five self-improvement baselines by 21.0 points, and improves every other base model by 17.2 to 40.3 points. Ablations attribute the gain to comparative proposal generation, the acceptance gate, and the hard regression budget rather than to skill writing itself, which without validation is no better than using no skills. The mechanism generalizes beyond the clinical domain, improving agents on three of four non-clinical environments and remaining flat only where the action space is open-ended. Frozen libraries transfer across models, where skills from a stronger model improve weaker executors beyond what they learn for themselves while the reverse does not, an asymmetry that no ungated baseline reproduces.

Recent progress in computer vision has produced a wide range of powerful specialized models for detection, segmentation, counting, and other visual tasks. However, these models are usually optimized for isolated task formulations, making it difficult to directly support general-purpose visual intelligence, especially when a task requires complex language understanding and dense small-object perception. In this paper, we propose VisHarness, a trainable visual agent that decouples high-level perception, reasoning, and decision-making from low-level task execution. Instead of training a model to solve a specific visual task, VisHarness learns to harness a set of carefully designed heterogeneous visual experts. This paradigm preserves the general intelligence of the agent while fully leveraging the precision advantages of specialized visual models in concrete visual tasks. With only lightweight training, VisHarness learns a generalizable visual expert-harnessing policy and can solve common fundamental vision tasks under various complex conditions through multi-turn interactions with visual expert models. To enable efficient on-policy reinforcement learning training in a live environment, we introduce dynamic visual memory archiving, which mitigates the rapidly accumulating visual-token overhead caused by multi-turn interactions with visual expert models. Experiments on four representative benchmarks covering reasoning segmentation, generalized referring segmentation, dense small-object detection, and referring counting demonstrate that VisHarness substantially outperforms existing general-purpose models and achieves competitive or superior performance compared with task-specific models.

HTML observations in LLM-based web agents are extremely long, and while many reduction methods have been proposed, it remains unclear which methods reduce overall agent latency while maintaining performance. The main obstacle is the high cost of end-to-end evaluation: in our experiments, evaluating 11 methods across 32 configurations on 33 tasks of WorkArena L1 required 232.4 cumulative hours. To address this, we propose a lightweight evaluation framework based on the Minimal Failure Set (MFS), the minimal set of HTML elements whose removal causes task failure. We define coverage as the fraction of instances in which a reduction method fully retains the MFS, which serves as a proxy metric that requires neither web access nor LLM inference. We validate that coverage strongly correlates with end-to-end success rate, with over 100$\times$ speedup in cumulative evaluation time on both benchmarks. Using this framework, we find that extractive HTML reduction methods require either high computation cost or domain-specific optimization to reduce agent latency while maintaining performance. Building on this, we optimize a pruning program on MFS training data, achieving 2.2$\times$ faster per-step latency on WorkArena L1 while retaining 84% of the original success rate, and 3.1$\times$ faster on WebLinx while retaining 89%.

Diffusion-based planning has achieved strong results in single-agent offline reinforcement learning, yet scaling to many-agent systems remains intractable due to the curse of dimensionality in the joint trajectory space. We introduce MF-Diffuser, a framework that lifts trajectory planning to the Wasserstein space of trajectory distributions, where the propagation of chaos ensures a small representative subset of agents captures the full population dynamics. Our approach features a value-weighted chaotic entropy objective that reconciles generative fidelity with return maximization, and a hierarchical coarse-to-fine strategy that progressively grows the agent population during denoising. We establish end-to-end suboptimality bounds with four interpretable terms, revealing that mean-field approximation error scales as $O(H^2/\sqrt{N})$ while offline distribution shift provably does not grow with population size $N$, and prove the generated policy is an approximate mean-field Nash equilibrium with explicit convergence guarantees. Experiments on three mean-field RL benchmarks — spanning stage games, sequential dynamics, and adversarial team competition — show MF-Diffuser achieves the best return in the majority of settings, with the largest gains on suboptimal offline data and at extreme scales ($N \geq 10^3$).

The impressive performance of generalist large language models (LLMs) such as GPT and Claude in healthcare raises a critical question: will domain-specific medical specialist models become obsolete? We argue that the future of medical artificial intelligence (AI) lies not in building monolithic medical foundation models, nor in replacing human expertise, but in orchestrating collaboration among generalist LLMs, domain-specific specialist models, and clinicians. We propose HetMedAgent, a heterogeneous medical multi-agent framework that enables conflict-aware evidence fusion, uncertainty-based clinician intervention triggering, and adaptive threshold calibration. Experiments on three real-world clinical decision-making tasks demonstrate that the synergy between generalist LLMs and domain-specific specialist models significantly outperforms using either type of model alone, validating the irreplaceable value of specialist models in modality-specific analysis. HetMedAgent represents a shift from building medical LLMs or foundation models to multi-agent collaboration, achieving a balance between general reasoning capabilities and domain-specific precision.

Web agents, which couple language models with browsing and tool-use capabilities, show promise as open web assistants. Yet progress is increasingly limited by the lack of scalable, process-level supervision. Existing benchmarks are largely manually constructed, providing only coarse start-goal annotations without intermediate trajectories, while recent automatic generation efforts remain expensive, biased, and shallow. These limitations prevent reliable training and evaluation of agents that must generalize to realistic, multi-hop, cross-page tasks. We introduce a scalable framework, GTA, that integrates crawling, retrieval-based seeding, in-context generation, and automated quality control to produce realistic tasks paired with executable trajectories. This design decouples crawling from generation for greater efficiency, grounds tasks in the site graph to enforce compositionality, and ensures dense supervision through deterministic replays and systematic validation. We instantiate the pipeline on over 50 websites covering e-commerce, government, forums, and news, with multilingual and multi-hop coverage. The resulting benchmark reveals a significant human-agent performance gap and enables detailed diagnostics. Our contributions are three-fold: (i) formalizing multi-hop web-agent task generation, (ii) proposing an efficient and validated pipeline for automatic data creation, and (iii) releasing a dynamic benchmark with reproducible evaluation.

Multimodal large language models are increasingly deployed as long-horizon agents, where memory must do more than recall: it must track an evolving world, revise what has gone stale, and surface the right evidence at decision time. Existing benchmarks measure recall over static dialogue, collapse memory into a single end-of-task accuracy, and reduce visual observations to captions, leaving us unable to localize failures to writing, maintenance, retrieval, or use. The rise of agent harnesses that author their own memory sharpens this gap, since we have no principled way to compare hand-designed pipelines with self-managing alternatives. To close these gaps, we formulate multimodal agent memory as an Action-World Interaction Loop with an observable four-stage lifecycle, and instantiate it in WorldMemArena: 400 multi-session multimodal tasks spanning Lifelong Evolution (evolving personal and task states) and Agentic Execution (memory from real observations, actions, and feedback), annotated with gold memory points, updates, distractors, and evidence chains for stage-level diagnosis. This enables the first head-to-head comparison of long-context, manually designed (RAG and external memory systems), and harness-based memory agents. Results show that: (1) better memory writing and storage do not guarantee better performance; (2) multimodal memory still struggles to fully use visual evidence; (3) systems are unstable across domains and degrade on realistic agentic trajectories; and (4) harness memory is more flexible but remains costly and less reliable.

Wearable devices enable continuous monitoring of physiological signals such as ECG and PPG, but existing mHealth systems are largely limited to task-specific prediction pipelines or reactive question answering over static summaries. They lack the ability to support temporal reasoning, persistent physiological context, and proactive monitoring over long-term signal streams. We propose VitalAgent, a tool-augmented agentic framework for ECG/PPG-based mHealth that supports both reactive question answering and proactive monitoring. VitalAgent is built on a longitudinal physiological memory and a tool-augmented reasoning interface that enables dynamic computation over raw signals. We further introduce VitalBench, a longitudinal physiological monitoring benchmark dataset comprising 1,862 QA pairs for reactive question answering and 90.2 hours of continuous ECG/PPG recordings for proactive monitoring, covering cardiac, physical activity, and stress-related tasks. Experiments demonstrate that VitalAgent achieves over 30% improvement over prompt-based and ReAct baselines in reactive evaluation and supports proactive alert monitoring over long-term physiological signals, highlighting the importance of dynamic tool use and long-term physiological monitoring.

Agent harnesses increasingly determine the performance of language-model systems by deciding how models call tools, receive feedback, verify intermediate states, store memory, and revise solutions. Yet current test-time scaling analyses often parameterize this process by raw expenditure — tokens, tool calls, operations, wall time, or cost — which does not distinguish useful feedback from redundant or unstable interaction. We introduce Effective Feedback Compute (EFC), a trace-level scaling coordinate that credits feedback only when it is informative, valid, non-redundant, and retained for subsequent decisions, and we normalize it by task demand when comparing tasks with different feedback requirements. Across synthetic controllable tasks, executable code tasks, real benchmark traces, held-out splits, and a prospective validation batch, EFC-based coordinates consistently predict failure rates better than raw-compute baselines and a strong multivariate SAS baseline. In controlled scaling, raw tokens and tool calls explain limited variation ($R^2=0.33$ and $0.42$), SAS reaches $0.88$, while Oracle-EFC and Estimated-EFC reach $0.94$ and Oracle-EFC/$D_{\mathrm{task}}$ reaches $0.99$. Matched-budget interventions show that improving feedback quality raises success from $0.27$ to $0.90$ while raw cost and tool calls are fixed. On mixed real traces, NRS-EFC/$D_{\mathrm{task}}$ reaches $R^2=0.92$ while raw compute has near-zero or negative fit, and it remains the best predictor in a prospective holdout ($R^2=0.85$). These results suggest that harness scaling is governed less by how much computation is spent than by how efficiently raw budget is converted into durable, task-sufficient feedback.

Consensus protocols form the backbone of distributed systems and blockchains, where implementation bugs can cause data corruption and financial losses. While LLM-based approaches show promise in code analysis, they struggle with deep protocol-level logic bugs involving complex state-dependent behaviors across multiple execution stages. We present Agora, a domain-aware multi-agent framework that integrates hypothesis-driven testing with LLM capabilities for systematic protocol verification. Agora employs specialized agents that collaboratively explore protocol state spaces, synthesize attack scenarios using domain-specific constraints, and validate findings through iterative refinement. This explicit role separation enables reasoning about global protocol invariants beyond single-function code analysis. We evaluate Agora on four consensus implementations (Raft, EPaxos, HotStuff, BullShark) using four state-of-the-art LLMs. Agora discovers 15 previously unknown protocol-level logic bugs that violate safety properties, while existing LLM-based agents fail to detect any such protocol-level logic bugs. Our results demonstrate that domain-aware multi-agent collaboration is essential for detecting deep logic bugs in complex protocols.

Large language model (LLM) agents increasingly leverage long term memory to support persistent and autonomous task execution. However, this capability also introduces a new attack surface: memory poisoning, where adversaries can inject malicious information to influence future behavior. Existing memory poisoning attacks often assume that injected content can be stored directly in memory, overlooking the selective extraction and rewriting stages in modern memory pipelines. This makes prior methods ineffective under realistic settings. In this paper, we propose MemPoison, a novel memory poisoning attack that bypasses selective memory mechanisms in LLM agents, where an attacker can inject triggerable backdoors into the agent's long-term memory through dialogue interactions, thereby misleading its subsequent responses. MemPoison introduces three key components: (i) a semantic relational bridge that binds the trigger and payload into a coherent statement to ensure they are extracted into memory together; (ii) entity masquerading that optimizes triggers to mimic named entities, resisting rewriting; and (iii) joint embedding optimization that shapes trigger-injected texts into a tight cluster in the embedding space while maintaining isolation from benign embeddings for stealth. Evaluations across different agent domains and memory mechanisms show MemPoison achieves attack success rates up to 0.95, outperforming existing baselines. Mechanistic analysis indicates that the attack exploits embedding-space anisotropy and shifts attention patterns, highlighting core vulnerabilities in selective memory systems. We evaluate multiple defense strategies and demonstrate their fundamental limitations in mitigating the attack.

AI agents augment large language models with external tools such as web retrieval, enabling grounded and up-to-date responses. However, incorporating external content into the generation pipeline can weaken the safety alignment mechanisms that govern model outputs. Prior work shows that enabling retrieval in agents increases compliance with harmful requests. We introduce AgentREVEAL, a diagnostic framework for analyzing retrieval-induced safety degradation in LLM agents. The framework examines two axes: how retrieval is integrated into the agent pipeline and the properties of the retrieved content. Along the integration axis, we find that binding tool invocation and response generation in a single step amplifies harmful outputs. Along the content axis, we uncover the Safe Source Paradox: even oppositional or safety-oriented sources, such as pages containing warnings or risk disclaimers, can increase harmful compliance by an average of 25% compared to the no-retrieval baseline. Finally, we show that relevance acts as a shared activation condition for both vulnerabilities. Similar patterns appear on frontier closed models, and harmful compliance remains elevated under several representative pipeline interventions, with some agents also entering this regime under autonomous retrieval. Because relevance is also what makes retrieval useful, these results expose a safety-utility trade-off for retrieval-enabled agents. We introduce HarmURLBench, a benchmark containing 1,405 real-world URLs paired with 320 harmful behaviors to support future evaluations.

Recent advances in mobile GUI agents have shown strong potential for automating mobile tasks, but most effective systems still depend on large vision-language models for screenshot understanding and long-horizon planning. Small GUI agents that can be deployed directly on mobile devices are more attractive for practical use, offering lower inference cost and better protection of sensitive on-device information. However, due to limited model capacity, such lightweight agents remain unreliable when planning and executing GUI tasks end-to-end from screenshots alone. We propose Knowledge-Oriented Behavior Exploration (UI-KOBE), a framework that improves lightweight mobile GUI agents with reusable app-specific graph knowledge. UI-KOBE first autonomously explores a mobile application and constructs an app knowledge graph, where nodes represent distinct UI states and edges represent executable transitions. At runtime, a lightweight GUI agent uses the graph as external guidance: given a user task and the current screenshot, it identifies the current graph node and selects among self-loop actions, neighboring transitions, task completion, or fallback free actions associated with that node. By supporting runtime decisions with app-specific graph guidance, UI-KOBE reduces the burden of end-to-end GUI planning and helps lightweight models perform mobile GUI tasks more effectively, offering a practical step toward efficient, interpretable, and privacy-conscious on-device GUI agents.

Memory-augmented LLM agents tackle complex long-horizon tasks by recursively summarizing interaction trajectories into compact memory. However, existing approaches typically train these memory policies using outcome-based reinforcement learning, failing to localize where intermediate memory quality degrades. As interactions unfold, ambiguous recursive summaries progressively discard task-relevant information and introduce semantic noise. This exacerbates belief deviation, obscuring the agent's estimate of the latent task state and ultimately derailing long-horizon reasoning. We therefore argue that memory optimization should focus not merely on trajectory-level success, but on the clarity of the belief induced by intermediate summaries. To this end, we introduce Belief Entropy, a self-supervised proxy that probes how uncertain the model remains about the latent task state given its current memory. Based on this proxy, we propose Metacognitive Memory Policy Optimization (MMPO). Instead of relying only on sparse outcome-based signals, MMPO provides fine-grained, memory-specific supervision via explicitly penalizing summaries that induce high epistemic uncertainty. Experiments show that MMPO consistently outperforms existing methods on diverse long-horizon tasks, maintaining 97.1% performance even when scaled to 1.75M-token contexts.

Large Language Model (LLM) search agents have shown strong promise for knowledge-intensive language tasks through multiple rounds of reasoning and information retrieval. Most existing systems access information using a retriever that takes a keyword or natural language query and returns a ranked list of documents using an index of pre-computed document representations. In this work, we explore a complementary perspective in which the search agent treats the corpus itself as the search environment and finds evidence by issuing executable shell commands. We introduce GrepSeek, an optimized direct corpus interaction (DCI) search agent that trains a compact search agent to find, filter, and compose evidence from large text corpora. To address the instability of learning behavior directly with reinforcement learning on large corpora, we propose a two-stage training pipeline. First, we construct a cold-start dataset using an answer-aware Tutor and answer-blind Planner to generate verified, causally grounded search trajectories. Second, we refine the initialized policy with Group Relative Policy Optimization (GRPO), allowing the agent to improve its task-oriented search behavior through direct interaction with the corpus. To make DCI practical at scale, we further use a semantics-preserving sharded-parallel execution engine that accelerates shell-based retrieval by up to $7.6\times$ while preserving byte-exact equivalence with sequential execution of the shell command. Experiments across seven open-domain question answering benchmarks show that GrepSeek achieves the strongest overall token-level $F_1$ and Exact Match. Our analysis also highlights the limitations of purely lexical interaction on queries with substantial surface-form variation, suggesting DCI as a practical and competitive method for search agents that can complement existing retrieval paradigms in the real world.

Marine lead (Pb) and its isotopes are critical tracers for ocean circulation and anthropogenic pollution, yet in-situ observations remain costly and sparse. While vast historical records exist, they lie buried within the unstructured content of academic papers, creating "data silos" inaccessible to comprehensive analysis. Manual extraction is unscalable, while general-purpose Large Language Models (LLMs) lack the necessary domain-specific knowledge, leading to hallucinations and scientifically invalid outputs. To address this, we introduce an expert-guided adaptation approach that enables LLMs to perform rigorous scientific data extraction without fine-tuning. We operationalize this approach through Compass, an LLM agent framework enhanced by a Knowledge Tree co-designed with marine scientists, which decomposes complex tasks into verifiable steps, guiding the agent's reasoning to ensure scientific validity. Deploying Compass across a corpus of over 230,000 relevant open-access papers, we successfully extract 3,751 previously unincorporated Pb records. This effort establishes the largest integrated marine Pb database to date. Beyond standard metrics, Compass demonstrates superior reliability through multi-layered validation, achieving 92% accuracy as confirmed through expert manual verification. The newly integrated data expand coverage in previously under-sampled regions such as the East China Sea and the Southern Ocean, providing an enriched data foundation for future scientific discoveries. We release an interactive visualization platform to facilitate open scientific access. Our work demonstrates that expert-guided agents can effectively bridge the gap between general-purpose LLMs and high-stakes scientific domains, enabling scalable data discovery in geosciences.

While LLM agents have demonstrated remarkable task-oriented abilities such as planning, reasoning, and action, few works have treated them as complete human personalities where emotional dimensions hold equal importance. In this paper, we introduce a novel benchmark to systematically assess whether LLM agents can simulate coherent, human-like psychology. Specifically, our benchmark constructs 11 diverse human characters grounded in orthogonal Big Five personality traits, with each profile deeply integrated with 1,000 structured autobiographical-style episodic memories distributed across theory-grounded developmental life stages. To rigorously evaluate the psychological manifestations of LLMs, we designed a curated suite of 64 decision-making scenarios, guided by the DIAMONDS taxonomy, a psychological framework that characterizes situations along eight dimensions: Duty, Intellect, Adversity, Mating, pOsitivity, Negativity, Deception, and Sociality. By subjecting agents to varying scenarios, the benchmark evaluates whether they can consolidate their innate personality traits and autobiographical memories to make behavioral decisions that are consistent with their specific psychological profiles. After systematic human validation and filtering, we obtained a benchmark consisting of 673 multiple-choice questions (MCQs). We believe this benchmark provides a principled and scalable testbed for studying human-like emotions, personality consistency, and value-consistent behavioural decision-making in LLM-based agents.

LLM-based agents have demonstrated strong capabilities in solving complex tasks through multi-step reasoning and tool use. However, existing evaluation protocols primarily focus on task success, overlooking a critical aspect of agent behavior: execution efficiency. In practice, agent trajectories often contain redundant steps that consume substantial resources while contributing little to task completion. In this work, we propose and formulate a new research area: redundant step detection for agent trajectories. To support this initiative, we introduce RedundancyBench, a new benchmark that contains diverse tasks with carefully annotated trajectories, where each step is labeled according to its contribution to task completion. Using RedundancyBench, we develop and evaluate 3 representative methods to answer whether a step within trajectory is redundant or necessary. Our results show that even the best-performing method achieves only 24.88% score in detecting redundant steps, while some methods perform worse than random guessing. These results highlight the task's complexity and the need for further research in this area. \footnote{Code and dataset in this paper are both available in \href{https://anonymous.4open.science/r/RedundancyBench}{https://anonymous.4open.science/r/RedundancyBench}.}