Agent 进展

Many LLM-as-a-judge evaluations score fixed outputs under a fixed task definition. Production multi-turn business agents instead require a maintained measurement system: correctness depends on business-specific facts and procedures, outcomes emerge across turns, and failures must be attributed to either agent capability or missing business knowledge before they are actionable. We present an integrated methodology spanning evaluation specification, modular LLM judges, intent-preserving user simulation, and human-in-the-loop governance. The specification defines conversation-level end states and actionable failure ownership. Atomic judges share versioned evidence and feed an explicit aggregation graph. The simulator is released only after task-preservation and stability checks. Independent human audits estimate measurement fidelity, renew tiered reference sets, and route disagreements to label correction, guideline revision, or judge improvement. Production studies show that system-level fidelity improved across repeated audits, that human reviewers and automated judges improved together under the shared feedback loop, and that their combined workflow had the strongest descriptive performance in both reported task-completion settings. Because the studies are observational and the human reference itself required revision, these findings demonstrate operational usefulness rather than causal or universal superiority. The contribution is a practical framework for making multi-turn agent measurement reliable, actionable, and maintainable as the evaluated system and its evidence evolve.

Long-horizon coding agents need timely corrections, yet feedback can be ineffective or even harmful when it misjudges ongoing work or fails to address the underlying problem. Existing critics focus on evaluating trajectories and generating feedback, but rarely track what happens after feedback is delivered. We present Opera, a verbal critic framework that treats each correction as a persistent note, followed until the diagnosed problem is resolved. Opera decides when to review through periodic and event-driven triggers, diagnoses issues with typed operators, audits feedback against visible evidence before delivery, and tracks the agent's subsequent actions to distinguish mere compliance from actual resolution. As a test-time critic, Opera improves the resolve rate of non-critic agents by up to 12.4, 15.0, and 8.9 percentage points on Terminal-Bench 2.1, a SWE-Bench Pro subset, and DeepSWE v1.1, respectively, across four policy models, and achieves the highest mean resolve rate among competitive critic baselines on all three benchmarks, and also improves policy models when the policy critiques itself. Beyond inference, Opera-guided rollouts provide approximately on-policy training data: fine-tuning Qwen3.5-9B on them improves its resolve rate on held-out SWE-Bench Pro repositories by 10.2 percentage points without a critic at inference time, matching fine-tuning on rollouts from a stronger model, while preserving its performance when switching harness, i.e., from Openhands to Terminus-2, which the latter substantially degrades. Our code is available at: https://github.com/dongyuanjushi/Opera.

AI agents now hold spend authority and settle payments without per-action human confirmation. The resulting loss is often not a security failure: a counterparty with the correct domain, the correct settlement address and a genuinely delivered service can charge more than it should, and no check keyed on identity will see it. We present three artefacts for measuring and reducing that loss. First, a taxonomy of agentic commerce fraud that separates five observation levels (agent reasoning, wire, settlement rail, counterparty, principal) from the request-level and history-level evidence available at each, and records which levels can observe which attacks. Second, Agentic Commerce Bench (ACB), a benchmark of twenty fraud classes generated from production aggregates, 1,647 catalogued service operations and 1,068 settlements, of which six involve a counterparty that is exactly who it claims to be. Third, gordonguard, an open-source detector stack and offline harness with which an operator can audit an agent configuration, replay hostile counterparties without an account, and run the same detectors inline. Calibrating to a stated false-positive budget on clean training traffic gives a 6.5% clean flag rate, replicated across three independent generations, and leaves eight of twenty classes no better than chance. On the four classes a reasoning layer can observe, a widely used agent security scanner run over its jailbreak-detection panel scores zero on all four, while correctly scoring 1.0 on a jailbreak supplied as a control. A measured median payment of $0.007 places a hard constraint on deployment: one human review costs 143 times the value of the payment it examines.

Memory is a core component of conversational agents, enabling coherent and context-aware behavior over long interactions. Recent approaches commonly rely on LLM-based memory construction, where raw interactions are rewritten into structured memory units and later retrieved via a RAG pipeline. While effective in controlled settings, we show that this paradigm breaks down in long-horizon, high-entropy conversations: memory construction becomes increasingly lossy and unstable as context length and information complexity grow, and incurs prohibitive cost due to repeated LLM invocation. To address these limitations, we propose Threader, a memory system that shifts the focus from memory construction to efficient, structure-aware access over raw interactions. Instead of rewriting interactions, Threader preserves them as first-class memory, organizes them into topic-coherent segments via lightweight incremental segmentation, and enables accurate retrieval through multi-view representation. At query time, it performs multi-signal retrieval that combines segment-level access with localized evidence matching, ensuring both completeness and coherence. Extensive experiments demonstrate that Threader consistently improves answer accuracy and evidence recall, while significantly reducing the memory construction overhead.

Large language model (LLM) agentic systems increasingly rely on models communicating with one another, yet existing uncertainty and multi-agent methods rarely estimate how a particular receiver will interpret a message before it is sent. This matters in heterogeneous systems, where capable receivers can reconstruct different tasks from the same message. We model this as a sender-receiver problem with a latent receiver type and define prospective interpretation risk (PIR): the probability that a receiver reconstructs a task other than intended. Rather than model an LLM's full input-output behaviour, we use black-box probes relating messages, intended tasks, and receiver-specific reconstructions, yielding scalable supervision while separating interpretation from downstream capability failure. Offline, heterogeneous frozen receivers provide supervision for receiver-conditioned risk and the effects of predefined mutable message features. At deployment, history induces a posterior over receiver types, guiding message revision and selection. We introduce value of interpretation information (VoII), querying for receiver information only when its expected communication benefit exceeds its cost. Our theory characterises when receiver information has decision value and bounds such queries. Empirically, interpretation-failure rates vary by 4-13x across receivers. Receiver information reduces PIR calibration error by 68% relative to a receiver-agnostic predictor, largely by correcting receiver-specific risk levels. PIR-guided revision reduces interpretation failure by 44% relative to the original message and 40% relative to a generic rewrite, mostly through a repair that helps every receiver. VoII outperforms information-gain and random querying at matched cost on the interpretation objective it optimises, lowering interpretation failure from 3.84% to 3.79% while querying 18.2% of episodes.

Tool-using language-model agents select and execute third-party artifacts. Different implementations can return the requested output while producing hidden execution effects that task-, attack-, or choice-based evaluations may miss. We study functional counterfeits: implementations that match benign alternatives on the requested output but add an effect forbidden by the task contract. We introduce SINGED (Source Integrity and the Nonidentifiability Gap in Execution Decisions for LLM Agents), a controlled benchmark covering five primary and two held-out task families. It varies displayed rank, evidence depth, decision policy, model release, and agent configuration, while task and process oracles verify the artifact and execution path. Across 7,549 audited trials, the randomized-rank study finds counterfeit execution in 45% (27/60) of rank-one trials and none at later ranks. Cross-candidate comparison eliminates shallow failures and reduces layered failures from 15.7% to 4.2%, but leaves dependency failures; its benefit is uncertain on unseen effects and public-package structures. Moreover, seven releases with no counterfeit executions when benign alternatives are available execute the counterfeit in 55/175 single-source cells after alternatives are removed. SINGED thus exposes a rank-, evidence-, and choice-sensitive outcome-to-execution gap: evaluation must connect correct outputs to execution paths.

Classical texts aligned with their translations support machine translation, retrieval and computational research, but evidence comparing alignment workflows is scattered. This study compares seven systems on 452 texts in Pali, Sanskrit, Mishnaic Hebrew and Tibetan, comprising 9,833 human-aligned units: four embedding pipelines, a direct LLM call, an autonomous agent, and the agent revised by an independent auditor. Generative workflows recover 93-94% of reference correspondences, against at most 77% for embeddings. A ceiling analysis shows that sentence boundaries make some references unrepresentable by the embedding pipelines. Reference recovery is similar across generative workflows: the agent's advantage is 0.5 percentage points (95% CI -0.02 to 1.17), and auditing adds no established benefit. Agents nevertheless produce structurally valid output for all 452 texts, against 437 for direct calls. A blinded three-LLM panel assesses every generative mismatch against the source and human reference. Most mismatches are labelled defensible editorial variation; consensus major-error labels cover only 0.06-0.14% of units. The panel labels significantly fewer residual defects for agents than direct calls (0.7% versus 1.4%), suggesting that reference recovery alone understates alignment quality. On ten long Pali discourses taken as published online, agents and audited agents raise recovery from the direct call's 71% to 84% and 92%. Identical reference-located chunks bring all three to 93%. Agents thus improve structural reliability and reduce judged defects on short passages, while their large recovery advantage on long documents disappears after chunking. In this setting, independent auditing offers little measurable additional benefit on prepared passages.

On-policy distillation (OPD) transfers the capabilities of a large language model to a smaller student by providing teacher supervision on the student's own rollouts. In long-horizon agentic tasks, however, uniform token-level matching can allocate supervision poorly: a large local discrepancy need not improve future behavior, while consequential guidance may be beyond the current student's reach or fail to persist without privileged input. We formulate long-horizon OPD as hierarchical supervision allocation and argue that productive guidance lies at the intersection of future utility and current learnability. Crucially, this intersection evolves as the student learns. Based on this principle, we propose LENS-OPD, a coarse-to-fine framework that organizes supervision through Locate, Validate, and Refine. Locate adapts trajectory exposure to the student's evolving competence and proposes a candidate decision for intervention. Validate tests whether teacher guidance at that decision improves the same student's subsequent behavior. Refine internalizes the beneficial guided behavior into the deployable policy and concentrates token-level supervision on decisive teacher-student conflicts within the validated turn. These stages are nested: each finer allocation is conditioned on the coarser decision, rather than being optimized as an independent importance score. Experiments across multiple long-horizon agent benchmarks and student-teacher configurations show that LENS-OPD consistently improves task performance over vanilla OPD and strong curriculum- and selection-based baselines. Our results suggest that effective long-horizon distillation requires teaching at the right depth, the right decision, and the right token.

Rapid progress in Large Language Models (LLMs) is saturating static benchmarks faster than they can be designed. While existing automated evolution frameworks attempt to generate harder questions by perturbing individual tasks, they remain constrained by rigid, hard-coded generation rules. Moving beyond the evolution of isolated tasks, we propose to optimize the benchmark generation workflow itself end to end with MetaBench-Harness, a dual-loop search framework. Specifically, the inner loop utilizes a benchmark harness to generate a new benchmark in each round, while the outer meta-harness orchestration layer iteratively refines and searches over harness implementations based on historical evolution trajectories. By applying MetaBench-Harness to the competitive programming CodeContests and Olympiad mathematics AIME-2024 datasets, we demonstrate that the evolved benchmarks are challenging and discriminative for frontier models. Trajectory and quality analyses verify that MetaBench-Harness enables multi-dimensional evolution, steadily improving evolution reasonableness, benchmark competency, and evaluator robustness across successive rounds. Furthermore, case studies reveal its effective utilization of diverse difficulty levers to reframe problems and elevate required capabilities. Ultimately, this work provides a solution to the pressing challenge of benchmark saturation.

Production scheduling requires realistic models that reflect operational constraints and efficient methods that balance competing goals. Putting these methods into use also requires data integration, model adaptation and specialist expertise. We present APEX, an extensible production scheduling framework built around a general model and hybrid multiobjective search. Agent assistance supports both scheduling and model refinement: agents prepare data and explore scenarios in natural language, while coding agents help implement and test new constraints and objectives. Shared construction and checking procedures connect these adaptations to the scheduling core. We benchmark eight APEX configurations against NSGA-II, SPEA2, MOEA/D and SMS-EMOA on 69 public job-shop, flexible job-shop and permutation flow-shop instances, assessing workload completion time (makespan), total job flowtime and computation time. Hybrid configurations achieve the best aggregate solution quality, although the leading method depends on the problem class and objective. A separate synthetic workflow study uses OpenAI's GPT-6-astra as an interaction layer between the human planner and the algorithmic core, testing rule additions, plan and objective changes, and what-if comparisons. All 24 sessions completed the requested changes and passed independent checks of saved models and schedules. A separate coding evaluation produced six native implementations of an additional objective or hard constraint through predefined extension hooks. All passed independent checks without modifying the core.

An LLM agent is defined by two things: the weights inside its model and the harness of components assembled around it. Harnesses are still handcrafted, and HarnessX, which evolves them automatically, starts each benchmark from a handcrafted harness, reports gains on the tasks it evolved on, and budgets 100 to 175 million meta-agent tokens per benchmark. We propose LiteEvo, a lightweight harness-evolution algorithm whose tool-free meta-agents mine agent trajectories for reusable components, curate them into a versioned library, and compose each round's harness from it, starting every benchmark from the same neutral harness and never naming the benchmark. Evolving on the graded tasks of five agentic benchmarks with a frozen Qwen3.5-9B, LiteEvo lifts pass@2 by 10.5 to 67.7pp and reaches comparable or higher pass@2 than a reproduction of HarnessX (71.0 against 67.3 on average) at 13.0 lower mean API cost. Harnesses evolved on train tasks keep their gains on unseen test tasks of four benchmarks, and LiteEvo also lifts Claude Code with Sonnet 4.6 by 1.2 to 71.4pp.

Code agents require long-horizon decision-making over complex interaction trajectories. However, existing reinforcement learning (RL) approaches typically optimize behavior at the token level, creating a mismatch between low-level generation and high-level behavioral reasoning. This limitation leads to inefficient exploration and weak credit assignment under sparse rewards. Moreover, while large-scale agent trajectories often contain recurring multi-step behavioral patterns, their noisy token-level representations hinder effective experience reuse. To address these challenges, we propose CodeSkill, a framework that adapts hierarchical latent skill modeling to the code agent domain. CodeSkill first leverages a teacher model to distill both successful and failed trajectories into multi-level textual abstractions. It then integrates temporal variational inference with reinforcement learning to map these discrete semantics into continuous latent variables, while an adaptive boundary mechanism dynamically gates skill transitions based on execution feedback. The learned skills are injected into a frozen LLM policy as latent semantic prefixes, enabling optimization in a compact semantic space rather than over raw token sequences. By shifting RL from token-level exploration to experience-level reasoning, CodeSkill improves optimization efficiency and long-horizon behavioral coherence. Extensive experiments demonstrate that CodeSkill achieves highly competitive performance against strong open-weight baselines across diverse general and industrial coding benchmarks. Furthermore, the learned skills exhibit strong transferability and robust cross-domain generalization, highlighting the effectiveness of explicit behavioral abstraction for scalable agentic code generation.

A tool-using language-model agent deployed over a long stream of tasks receives no reward, so it cannot tell whether it succeeded, cannot safely retry, and cannot label the experience it needs to improve. We present SelfSuite, in which the agent's own base model, given only the world's public materials, designs a small evaluation suite of weighted judges and grounded per-task briefs, freezes it, and uses it to gate a keep-best retry and to label a typed, outcome-tracked memory. On matched five-repeat benchmarks over tau2-bench and AppWorld, SelfSuite scores above the plain agent without any labels, matches methods given ten expert labels on tau2-bench, and trails Agentic Context Engineering (ACE) on AppWorld, where code execution gives a direct success signal. In an ablation campaign run on the same tasks, it is above label-free ACE in every repeat, and the gated second attempt is the only component whose removal hurts in every repeat. We also simulate a subject-matter expert who grades ten onboarding tasks per world. Using those labels to calibrate SelfSuite's evaluator gives a small, consistent gain, and using them to warm up ACE's memory lifts ACE to tie calibrated SelfSuite. A single-run study on a second model family shows the same ordering.

An agent harness controls how a language model accesses information, uses tools, preserves memory, and checks its work. Improving this software is costly when each evaluation requires a long interaction with an environment. We introduce Vestrum, a framework that turns failures in execution traces into scoped harness changes without training the task model. Its organizing overhypothesis is that tasks of a shared kind may exhibit recurring failures whose remedies transfer within that kind. Vestrum expresses failures as recognizable classes, proposes changes across verification, retrieval, decomposition, and knowledge synthesis, and screens their scope before evaluating them as a bundle. A persistent lessons file informs subsequent proposals. Across five settings and two baseline harnesses, the frozen harnesses improve held-out performance: UltraHorizon rises from 47.6 to 59.8 over GAM, Terminal-Bench 4 Hard from 63.7% to 70.3% of checks passed over Claude Code on eight held-out tasks at 1.03x test cost, and cell-type annotation agreement from 67.5% to 77.8% on held-out sections of one slide, alongside gains on LoCoMo and AMA-Bench. Across our searches, verification grounded in evidence helped both intermediate steps and final answers, at lower cost at intermediate steps, while critics asked to rebuild finished answers broke more than they repaired. On the three memory benchmarks, Vestrum also scores above the evaluated GEPA configurations in every paired evaluation.

Personal Graphical User Interface (GUI) agents rely on interaction history to infer what a user wants from ambiguous instructions and to anticipate recurring routines. Existing approaches retrieve task-relevant history and append it to the policy's context, implicitly assuming that experience relevant to a task remains useful for each decision within it. We find that this help is largely spent at the first decision: retrieved history strongly improves the opening step of an episode, yet provides little sustained benefit over the remaining 90% of steps, and offers weak guidance on whether a proactive suggestion is warranted. A relevant record may tell the agent where to begin, but not which past action applies to the current screen or whether a routine is due now. The underlying issue is that relevance does not imply applicability}: relevance is determined at the task level, whereas applicability depends on the situation at decision time. We therefore recast personalization as experience activation and introduce ExpActivator, a training-free framework that activates only the experience applicable to the current situation. During execution, ExpActivator matches each new screen to historical states in the frozen GUI backbone's latent space and supplies the corresponding action as a reference. Before execution, it activates a recurring intent only when the current time and scenario provide sufficient support, and otherwise abstains. Across four GUI backbones, ExpActivator improves within-trajectory step success by 28% on average, achieves the best personalized execution on every backbone while using about one-fifth as many history tokens, and reaches approximately 2.3$\times$ the Matthews correlation coefficient of the strongest proactive baseline. Experience pays where it is activated, not where it is appended.

Browser-use agents require seamless alignment between structured web metadata and visual information, while preserving relevant context across long interactions. Existing interfaces often rely on either screenshot-level action prediction or static Set-of-Marks overlays, leaving the model to resolve dense DOM-pixel alignment before every operation. We introduce Probe to Act (P2A), an active probing framework for the browser-agent loop that moves this alignment into decision time. P2A addresses an asymmetric bridge between symbolic DOM hypotheses and screenshot layout by rendering on-demand symbolic DOM structure back into pixels. Before committing a state-changing browser operation, the agent can issue lightweight probes to translate DOM handles into pixel evidence, map screen regions back to DOM candidates, register visual-only targets, and commit verified notes. These interleaved processes naturally produce evidence-based memory: only probed, acted-on, or explicitly committed observations are kept across steps, preserving only decision-critical evidence in long-horizon contexts. P2A can be used as a prompting strategy for proprietary models under the standard DOM+SoM interface, and can be distilled into open-weight models through cold-start synthesis and self-bootstrapped SFT. Across three browser-use benchmarks, P2A shows clear gains on task success rate for both proprietary and fine-tuned models; on VisualWebArena, for example, it improves Gemini-3-Pro from 54.1% to 61.2% and Qwen3-VL-8B from 24.6% to 32.9%, while matching the costly full-observation history ($\sim$3$\times$) at only $\sim$1.2$\times$ the peak retained input context of action-only history.

LLM agents increasingly take consequential actions through interactions with users, policies, and external tools. Auditing these agents requires automated attribution of realized actions to their historical basis. However, existing attribution formulations do not provide question-specific traces for diverse auditing objectives. Additionally, when access to the acting model is limited (e.g., in API-only deployments), applicable methods commonly rely on costly input perturbations or external LLM analysis of complete trajectories. We therefore formulate query-conditioned agent action attribution, a new task that takes a natural-language auditing query as input and recovers the source and ordered intermediate evidence for the query-specified aspect of an action. We instantiate this task with $A^3Bench$, a benchmark comprising 1,396 auditing queries across policy basis, parameter provenance, failure propagation, and unsafe-behavior tracing. To enable efficient, query-specific attribution, we use small open-weight models as attribution proposers that combine query-conditioned gradient saliency with query-semantic relevance to rank history units. Our proposer consistently achieves stronger source and evidence rankings at lower inference cost than open-weight baselines, improving source MRR by up to 40.9% and evidence MAP by 42.1% with only two forward passes and one backward pass. Controlled evaluations confirm that our proposer improves attribution specificity by adapting its rankings to fine-grained changes in the auditing query. Building on a proposer ensemble, our end-to-end system surpasses the strongest frontier-model baseline in source accuracy (64.5% vs.\ 60.4%) while reducing empirical deployment latency by 29.9% relative to the fastest frontier API baseline. Code and data will be released after the initial review period following final validation and cleanup.

Coding agents now sustain hours-long, tool-driven loops, yet their ability to carry long-horizon analog and mixed-signal circuits to electrical specification remains unmeasured. We introduce Analog Design Bench, a long-horizon agentic benchmark of 50 transistor-level design tasks contributed by 17 chip designers. Agents work with an open-source simulator, while an isolated verifier evaluates the submitted circuit using specification-based electrical tests. We evaluate 15 agent configurations across 2,250 two-hour attempts and observe full-specification pass rates from 8.0% to 78.0%. Coding-benchmark performance correlates with analog results but leaves much of the performance spread unexplained. Our failure analysis shows that most unsuccessful submissions have no recorded legality rejection but fail electrical acceptance, identifying electrical closure as the dominant endpoint challenge. We test time, reasoning effort, agent harness, and supplied design knowledge as interventions. Longer budgets and higher reasoning effort improve performance, while general skill documents provide little benefit and sometimes reduce performance. Supplying a task-matched reference topology, an idealized form of circuit-IP retrieval, raises DeepSeek V4 Pro by 18.7 percentage points and mainly accelerates GPT-5.6 Sol.

Automatically generated environments provide a scalable source of interaction data for training general agents. However, existing approaches mainly generate tasks within a single environment, while real-world workflows require agents to connect information and actions across multiple services. We introduce Compositional Environment Scaling (CompoWorld), which expands the task space by composing a finite library of reusable services. Coding agents turn tool specifications into verified services with typed states and shared interfaces, while a world model handles tools that cannot be reliably implemented. A random-walk procedure connects services through dependency graphs, enabling the generation and verification of tasks that require information to flow across services. Verified trajectories support supervised fine-tuning (SFT), while our Completion-Focused Rubric Reward guides reinforcement learning (RL) toward full task completion by emphasizing criteria with lower pass rates within each rollout group. We construct 448 services exposing 10,130 tools and use 3K SFT trajectories and 1K RL tasks to train Qwen3.6-35B-A3B. Experimental results show that CompoWorld improves on its backbone by 9.17 points on average across eight benchmarks. On AutomationBench, it surpasses frontier models such as Claude Opus 4.6 and leads all compared agent-specialized 35B-A3B models.

Large language model (LLM) agents increasingly undertake extreme-long (xlong) horizon tasks, where a single execution can span hours, hundreds of model--environment interactions, and nearly 1M tokens per rollout. Applying online reinforcement learning (RL) to such executions poses two fundamental challenges: (1) severe execution variance and prolonged rollout delays cause massive GPU idling; and (2) complex non-linear branching generates massive trajectory redundancy, crippling training efficiency. To address these, we presents QwenGyre, an end-to-end framework for xlong-horizon online RL. QwenGyre elastically reallocates GPUs between rollout and training without interrupting live executions, while its trajectory processor reconstructs branching histories, scores partial progress, and deduplicates redundant paths to bound training costs. Scaled to our flagship model, Qwen~3.8 2.4T, with 700K tokens per rollout, QwenGyre yields a 6.0% absolute gain on NL2RepoBench (52.5% $\to$ 58.5%) in 48 steps. Across our evaluations on diverse domains of training datasets, QwenGyre delivers up to $1.85\times$ and $1.78\times$ speedups over Colocate and Async, respectively.

Equity-relevant news evolves through temporally dependent corporate events, making historical information useful only when event continuity, information availability, and transition reliability are modeled. Existing LLM-based financial agents incorporate historical evidence, yet they provide limited support for preserving issuer-specific chronology under point-in-time constraints and for identifying when historical transitions contribute information beyond the current forecast. We present RICE-Alpha (Reliability-Informed Correction with Event Graphs), a point-in-time stock-scoring framework that separates a history-aware multi-view Base Alpha from a reliability-calibrated residual correction derived from historical event continuation. A Multi-Tier Memory Layer grounds news interpretation in temporally eligible issuer-specific history, while a Typed Event Agent constructs event states whose successor relations are formed within issuers and pooled across firms only after valid local pairing. Matured transitions are calibrated by their empirical reliability, and the resulting graph signal is residualized against the Base Alpha and technical view to obtain the RICE Delta. On daily Nasdaq-100 and Hang Seng Index panels from 2024 to 2026, RICE-Alpha achieves the strongest results among the evaluated LLM-based agents and momentum across four predictive and four portfolio-level metrics. Its ICIR more than doubles that of the strongest baseline, while net Sharpe ratios reach 1.656 and 1.725 in the U.S. and Hong Kong, respectively. U.S. ablations further show significant reductions in IC and RankIC after Holm adjustment when major components are removed. These results indicate that historical event continuation adds incremental information when it is temporally grounded, reliability-calibrated, and introduced as a residual correction to a multi-view forecast.

Large language models (LLMs) are increasingly used to drive scientific discovery. Understanding how LLMs make decisions from new data and memory of the literature is vital before trusting them to design experiments under tight experimental budgets. However, their decision strategies are invisible in the current performance scores used to evaluate research agents. Here, we benchmark five frontier LLMs in a Bayesian Optimization setting against published statistical baselines on seven combinatorial datasets spanning protein engineering, reaction optimization, molecular design, peptide self-assembly, and catalysis. Performance is paired with direct measurements of model beliefs and actions, enabling highly resolved behavior analysis. A prompt ablation that progressively strips context separates memorization from chemical reasoning and from bare categorical optimization. Prior chemical knowledge helps in expectation, but with high variance and occasionally even harms performance. No configuration tested decisively beats a mean statistical baseline across domains. Belief-movement and Martingale diagnostics, corrected here for a measurement-noise bias that mislabels rational agents as irrational, show that models overreact to incoming data rather than entrenching on their priors in the contexts studied here. Interestingly, while LLM actions are exploitative, models sincerely intend to explore and consistently act on that intent. This failure is a competence gap arising from context-stickiness. Removing in-context history restores exploration, indicating that priors and data must be decoupled to achieve effective LLM-driven discovery.

Reported improvements from tools and reusable skills in large language model agents refer to different comparisons. This critical narrative review examines what these evaluations estimate and which conclusions their designs support. The review checks the roles of one hundred cited papers and extracts focal evaluation designs in detail from thirty-five studies. Targeted readings of thirty-five additional published or accepted studies broaden coverage of tool creation, memory, interactive benchmarks, reliability, and risk. Designs are characterized by treatment contrast, target population, outcome, budget constraint, summary measure, and identification assumptions. Analytic decompositions and counterexamples show that pairing runs on the same task does not itself identify an invocation effect when evaluation conditions on a trigger within the treated run. Paired gain and regression counts describe discordance under the coupling protocol rather than the share of tasks whose expected outcomes worsen. Total effects of deploying a module answer a different question from efficiency under a common budget. Comparisons across studies distinguish curated skill provision from retriever replacement, task populations from triggered subsets, and preparation costs from marginal usage costs. Publication status and reading depth are recorded. The review provides a methodological synthesis and a reporting checklist to help align claims about tools and skills with the comparisons their evaluation designs support.

Large Language Models (LLMs) enable us to better understand text documents, including PDFs and Word documents. However, LLMs, as well as more modern LLM agents, i.e., those with tool-calling abilities, typically treat such documents as plain text, ignoring the fact that they are often organized hierarchically into sections and subsections. Extracting this structure, while difficult, can improve efficiency and effectiveness for agents (and humans)---since only sections relevant to a given task need to be processed. Unfortunately, prior work on structure extraction provides no formal guarantees on how well the inferred structure matches the true one. Instead, we target a robust and compact variant that is feasible to infer and useful in practice. Robustness ensures that the text under each subsection header is a superset of the text under the same header in the true structure. Compactness seeks to minimize this superset, reducing agentic cost (or human cognitive load). We propose SHED, a two-stage workflow for inferring a robust and compact structure. The first stage is pluggable with an infinite family of approaches, each guaranteeing robustness for a specific document class. We theoretically characterize the document space using these classes and their hierarchical relationships. Empirically, SHED improves F-1 scores (measuring the robustness--compactness trade-off) by 13%--68% over non-LLM baselines and 9%--15% over expensive LLM-based approaches. Finally, we show how SHED-inferred structures are valuable for agentic document analysis: agents using SHED outperform baselines, achieving 3%--23% higher accuracy while being up to 10x cheaper.

Language model agents can now perform sophisticated sequences of actions via tools and harnesses, which has increased the scope of the damage they can cause. Guard models, however, are mainly built for content moderation and thus are not well-suited to detecting this agentic risk. To address this, we proceed by first conducting a representational analysis, then use the resulting insights to build a solution. In our analysis, we focus on two types of trajectory-level agentic harms: harmful content, which is expressed directly, and unsafe tool use, which depends on whether an action is consistent with the interaction that produced it. We investigate how open-source guard models represent these two types of harm and find that they are linearly readable inside the model, even though guard models predict no better than chance on pairs that differ only in the called tool's schema. The two harm types also follow nearly orthogonal internal directions, and neither reliably serves as a proxy for the other. These results motivate reading trajectory safety directly from internal states. We introduce TACIT, a readout of a frozen backbone's internal states that decodes no tokens. Trained on six trajectory-safety benchmarks, a linear probe raises mean macro-F1 from 62.3 for the strongest open guard to 80.7, and refined readouts reach 86.2. With each benchmark held out of training entirely, the refined readouts still lead the strongest guard (65.7 vs. 61.1). With the same backbone, training data and test split, the frozen readout is on par with full safety fine-tuning, and it improves the fine-tuned model further when applied on top. The probe trains about one millionth as many parameters as full fine-tuning in about a sixth of the time, and TACIT has the lowest latency of the guards we evaluate.

Language agents choose what to remember about a person and how to use that memory. We separate these choices when predicting a person's unseen answer to an interview question. On 1,768 tasks from 188 people, a concrete memory from a verified interview prefix outscores a trait description by 0.0158 (95% interval [0.0044, 0.0271]). Crossing both memories with one-shot generation and three-answer fusion, fusion lowers concrete-memory scores by 0.0123 ([-0.0189, -0.0056]); prompted and trained selectors do not detectably beat a random candidate, and one call on the longer, unrewritten record outscores every memory condition. At matched context budgets, OwnWords, one call on the person's BM25-ranked sentences, outperforms the written memory on 500 people outside the benchmark (+0.0127, [+0.0037, +0.0217]; an earlier held-out test was inconclusive) and across four budgets on 300 people (mean +0.0218, [+0.0138, +0.0298]), but does not detectably outperform recency truncation; these comparisons do not isolate verbatim wording. On a survey benchmark it predicts ordinal answers more closely than the memory but is not more accurate on exact choices (less accurate in one of two screener samples). Interview scores use a model-based content rubric without human ratings, and all benchmark and some confirmation participants were seen during development.

Language model agents are increasingly deployed in open-world tool environments, which require balancing exploring unknown capabilities and exploiting known ones. Existing methods face a performance-efficiency tradeoff: they either rigidly decouple exploration and execution or interleave them without coordination. We argue that the key lies not in whether to decouple or interleave them, but in how to coordinate them across granularities. We introduce ParaAct, a structured parallel-action loop that combines phase-level Exploration $\rightleftharpoons$ Execution with action-level parallelism. To learn this loop, ParaAgent combines multi-agent cold-start demonstrations with reinforcement learning under multi-level advantage decoupling, making planning structure explicit and supervising it with step-, phase-, and trajectory-level rewards. Learning is supported by our ToolEnv, a scalable simulator grounded in 50,011 realistic tool interfaces. On two open-world tool benchmarks, ParaAgent-4B achieves the best average success among all baselines, including GPT-4.1 systems, with the largest gains on multi-tool tasks. Behavioral analyses show that these gains stem from this action organization, highlighting its importance for capable and efficient open-world agents.

LLM agents increasingly solve long-horizon tasks by autonomously interacting with their environment. In doing so, their strategies rely on assumptions about that environment: which resources and tools exist, where they are located, and how they behave. When these assumptions no longer hold, reliable agents must detect the change and adapt while pursuing the same goal. We study this adaptation capability through environmental novelty: a change that keeps the task objective fixed while invalidating an assumption underlying an otherwise successful trajectory. We introduce AGNI, an automated pipeline that extracts trajectory-relevant assumptions, injects targeted environmental changes, and validates that the resulting novel tasks remain solvable. Across three terminal benchmarks, AGNI produces diverse novelties spanning resources, interfaces, constraints, and execution semantics. Evaluating multiple LLM agents reveals a substantial adaptation gap between base and novel tasks. Trajectory analysis suggests that agents often encounter evidence of the change but fail to diagnose its cause and revise their strategy. Finally, post-training for environmental novelty improves adaptation to held-out novel tasks while also improving performance on base tasks. Our results highlight a gap between task competence and adaptive capability and motivate environmental variation as a core dimension of agent training and evaluation.

Autonomous large language model (LLM) agents operating in multi-product markets must make sequential decisions under information asymmetry and resource constraints. We develop a machine learning approach for training such agents to act effectively as sellers in a multi-item bargaining environment, where a seller concurrently negotiates a catalog of substitutable assets across a pool of independent buyers. Buyers hold private, heterogeneous valuations across products, and each can purchase at most one item. Facing limits on total communication turns, the seller must dynamically match buyers with the most profitable products considering their private valuations, while strategically allocating its limited interaction budget toward combinations of greater potential value. We formalize this problem as a Partially Observable Markov Decision Process using a structured, four-part message protocol that maps natural language into a parsable and regulated decision space. Using this formalization, we design a post-training method using Reinforcement Learning from Verifiable Rewards (RLVR). To evaluate this framework, we construct a multidimensional metric suite that quantifies constraint adherence, seller surplus extraction, and allocation quality. Our trained seller agent learns to match limited inventory to buyers more effectively, matching or outperforming trillion-parameter frontier models in both seller surplus extraction and buyer-product allocation quality. Finally, these learned strategies generalize robustly to unseen market structures, correlated valuation distributions, and price ranges not encountered during training.

World models improve sample efficiency by training policies on imagined trajectories, but their usefulness depends on learning representations that capture the information needed for future control. We study whether self-supervised joint-embedding prediction (JEPA) can provide this learning signal for multi-agent reinforcement learning. We introduce MA-JEPA, a stochastic world model that replaces observation reconstruction with prediction of target representations, enabling model-based multi-agent reinforcement learning with centralized training and decentralized execution. A categorical latent state and a causal Transformer are trained with posterior and action-conditioned dynamics prediction objectives and are then used for actor-critic learning from latent imagination. A training-only joint predictor conditions on all agents' local states and actions to predict each agent's next local observation embedding. These predictions are passed through the same local posterior used during real interaction with a centralized critic that is used only for value learning, with execution remaining decentralized. Our experiments show that this architecture performs strongly on SMAC, matching or exceeding the strongest reported comparator mean win rate on four of eight evaluated maps.