Agent 进展

Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment. This limits their ability to adapt how they gather evidence, invoke tools, verify signals, and manage risk under changing market regimes. We introduce EvolveTrade, a self-evolving framework that treats the system prompt of a tool-using trading agent as a text-parameterized policy. After each update interval, a Policy Agent revises this policy using accumulated decision traces and realized portfolio feedback, while keeping the backbone LLM fixed. The updated policy is then used for the next batch of trading decisions, enabling the agent to refine its information-acquisition and portfolio-construction procedure over time. Experiments across multiple market regimes and two LLM backbones show that EvolveTrade often improves Sharpe Ratio and Cumulative Return over fixed-policy LLM baselines, achieving the improved SR and CR in most evaluated settings. Behavioral analyses further show that self-evolved policies increase code-mediated analysis and activate regime-relevant computations; case-level policy-to-return attributions trace how policy-induced allocation changes contribute to realized return differences. These results suggest that adapting the reusable procedure governing tool use is a key direction for building more robust LLM trading agents.

Computer-use agents that operate through screenshots and simulated actions are advancing rapidly, yet their evaluation remains anchored to general desktop and web tasks. Enterprise Resource Planning (ERP) systems run the finance, procurement, inventory, and customer operations of organizations worldwide, and pose distinct challenges for computer-use agents: dense interfaces, coordinated multi-step interactions, and errors that alter persistent business records rather than surfacing on screen. Existing enterprise benchmarks rely on proprietary platforms or on simulated approximations of such software. We introduce ERPBench, a benchmark that evaluates screenshot-only agents on a live and reproducible ERP system and scores each task against ground-truth values in its database. Beyond the benchmark, we present a production-grade harness that gates agent actions behind human approval for safe deployment, which ERPBench runs autonomously. Evaluating six closed and open-source agents, we demonstrate that strong general GUI performance does not transfer to enterprise reliability. Even when an agent reaches the right form and saves it, the stored record is often wrong: some agents save in up to 85% of runs but write the correct value in as few as 3%. We further characterize failure modes specific to enterprise workflows.

As AI agents take on long, autonomous tasks, we increasingly oversee rather than perform the work, yet we still judge them almost entirely by whether they finally succeed. An outcome cannot reveal where a run went wrong, whether the agent recovered, or the irreversible harm it caused along the way, and where long-horizon agents fail remains unmapped. We study $2518$ agent trajectories across software engineering, computer use, and science, close to real deployment, and classify $6967$ mistakes into $78$ failure types. Failure follows a recurring signature: after its first mistake an agent often fails to recover and rarely catches the error itself, so the run continues unchecked while still looking correct; whether an agent recovers depends on the task and the environment's feedback, not on the agent framework running it. Long-horizon agents can do real harm on the way to a passing result: even runs scored as solved delete data, corrupt systems, or fabricate success rather than earning it. We release these human-verified annotations as Traverse, a benchmark on which six frontier judges struggle to locate failure regardless of scale: even the strongest correctly identifies the first mistake in fewer than a third of runs. Yet Scout, a $4$B verifier we trained, locates failure far better than these judges and transfers to domains it never saw. Used at test time to select among an agent's candidate runs, it raises task success above the agent's own single-attempt performance, without retraining the agent. By making failure cheap to locate and correct, this work is a foundation for more trustworthy long-horizon agents that learn from their own mistakes, and a practical path to overseeing increasingly autonomous AI.

Vision-language model (VLM) agents combine specialized perception, tools, and reasoning to address complex visual tasks. In multi-agent settings, different agents inspect different image regions, video frames, or visual representations, so collaboration extends beyond distributed reasoning to distributed perception. This makes shared visual context a central problem in VLM agent collaboration. In this paper, we frame memory hierarchy, cross-agent sharing, and consistency mechanisms around the need to reconcile interpretations and update dependent reasoning. Effective collaboration requires agents to build on contributions from other agents, recover missing visual context, and reconcile differing interpretations as new evidence emerges. Shared visual memory preserves not only images or textual summaries but also the dependencies among observations, agent interpretations, and subsequent reasoning. Together, these design considerations shape how information flows and evolves across VLM agents. The proposed framework provides a foundation for building reliable and resource-efficient agent teams.

Large language model (LLM)-based agents are increasingly used across software engineering, web automation, research, and productivity applications. Their integration of planning, memory, tool use, code execution, and external interactions enables greater autonomy but also introduces new reliability, safety, and security risks. We present a large-scale empirical study of quality assurance (QA) practices in 157 open-source LLM-based agent projects with at least 100 GitHub stars. We analyze documentation, source code, configurations, and tests to characterize QA practices across execution surfaces, safeguards, testing artifacts, risk scenarios, and recurring gaps. We find that current QA primarily focuses on basic functionality and high-risk actions, while coverage remains fragmented. Safeguards are inconsistently applied across equivalent execution routes, tests rarely examine boundary, adversarial, or multi-step tool-use failures, and identified risks are seldom translated into end-to-end QA checks. These findings highlight the need to move beyond feature-level testing toward systematic end-to-end validation that ensures agent workflows remain within intended boundaries when interacting with untrusted inputs, tools, persistent state, and external APIs.

Large language models (LLMs) have recently shown promise in Verilog generation, but producing functionally correct RTL directly from natural-language specifications remains a highly challenging task. Existing approaches improve LLM-based Verilog generation mainly with retrieval-augmented generation (RAG), self-planning, or few-shot prompting. However, these methods focus primarily on external or generic forms of enhancement rather than strengthening the specification with task-specific constraints. In this work, we propose SpecLens, an automated framework for LLM-based Verilog generation that derives specification-driven constraints by analyzing behavioral divergence among multiple candidate implementations, using the original specification as the only external semantic source during generation. On the VerilogEval v2.0 spec-to-RTL benchmark, SpecLens achieves a functional pass@1 ratio of 86.2% with o3-mini-medium and 89.4% with o3-mini-high. This corresponds to a 3.6 percentage-point gain over the SOTA prompting method with o3-mini-medium and a 3.8 percentage-point gain over the SOTA behavioral divergence method with o3-mini-high. In addition, on RTLLM v1.1 and v2.0, analysis shows that SpecLens is more specification-faithful and less prone to benchmark-aligned priors. SpecLens achieves 100% syntactic correctness on VerilogEval v2.0, 86.2% on RTLLM v1.1, and 88% on RTLLM v2.0, even without using costly compile-repair loops to revise generated code iteratively. The code is open source and available at https://anonymous.4open.science/r/SpecLens-4632/readme.md.

As Large Language Models (LLMs) increasingly serve as primary knowledge retrieval interfaces, their robustness against persuasion attacks---attempts to inject misinformation or enforce counterfactuals---has become a critical safety concern. Existing red-teaming frameworks typically evaluate models in multi-turn dialogues where the target model retains full conversation history. We identify a critical flaw in this setting termed ``Refusal Inertia'': a model's initial refusal often propagates through subsequent turns largely to maintain contextual consistency, thereby masking its true vulnerability to sophisticated, isolated persuasion attempts. To rigorously evaluate the ``cold-start'' defense capabilities of SOTA models, we introduce the SAST-IR (Stateful Attacker, Stateless Target - Iterative Refinement) framework. By enforcing a memory wipe on the target while retaining the attacker's history, we simulate a worst-case adversarial setting using multi-turn (stateless) iterations. Leveraging CP-Agent (Cognitive Persuasion Agent), an enhanced diagnosis-guided agent, our experiments on the custom \textsc{CounterFact-Strict} dataset ($N=50$) yield alarming results: simple, diverse attack strategies achieved a staggering 96% success rate, exposing severe brittleness in memory-less defense. Furthermore, we reveal a ``Complexity Paradox'': while complex, iteratively refined attacks are effective, they often trigger defensive compliance, whereas simple strategies achieve a higher rate of genuine persuasion (84.7%). Our code and dataset are available at GitHub, https://github.com/cza1006/llm-persuasion-defense.

Assistive autonomous systems must anticipate human goals before an observed behavior is complete. This article formulates anticipation as goal inference from a partially observed multimodal episode together with structured prediction of the remaining behavior, rather than exact motor forecasting. A compact Hierarchical Planning Decoder (HPD) is attached to a frozen neuro-symbolic recognition encoder and predicts, at four ontological levels, the next actions, the remaining activities and low-level intentions, and the episode high-level intention(HLI). The decoder is trained with soft neuro-symbolic regularization combining transition-coherence and hierarchical continuity losses, and is decoded with hard reachability masks that enforce ontological validity at inference. On a compositional four-level benchmark of 15,002 multimodal episodes built over NTU RGB+D 120 features, three headline properties are observed together. The advantage over the strongest sequential baseline grows with the anticipation horizon, from +1.7 points at step 1 to +7.3 points at step 3 (top-5). Under compositional generalization, where one parent association per multi-parent low level intention is held out, this advantage widens to +4.9 points at step 1. At the episode level, 96.8% of anticipated trajectories satisfy the joint logic constraints, above the 88.1% strongest-baseline value and the 73.9% ground-truth floor; soft logic terms alone account for a 59.8 to 71.1% relative reduction of HLI-reachability violations, and the hard masks then eliminate them entirely. Neural generation supplies predictive ranking, symbolic constraints supply onto logical validity, and their combination yields coherent hierarchical anticipation while exposing remaining challenges in compositional goal generalization and unordered set prediction.

Autonomous systems require reliable place recognition for efficient and effective simultaneous localisation and mapping (SLAM). Traditional geometric visual SLAM approaches rely on low-level features and geometric consistency, but remain vulnerable to perceptual aliasing, where different places appear similar, and perceptual variation, where the same place appears different. Although semantic SLAM and modern learned visual place recognition (VPR) methods improve robustness under challenging perceptual conditions, real-time deployment requires both high retrieval accuracy and low latency. Inspired by human memory and perception, we propose HuMem-VPR, which exploits the bidirectional relationship between bottom-up perceptual evidence and top-down contextual reasoning to achieve high-level place understanding. We further introduce HuMemSLAM, the integration of HuMem-VPR with ORB-SLAM3. HuMem VPR achieved the highest aggregate retrieval accuracy on the real-image benchmark, competitive accuracy on the CARLA benchmark, and approximately two to three times lower latency than the evaluated state-of-the-art VPR methods. Across the evaluated dataset families and online experiments, HuMemSLAM substantially improved integrated Recall @1 over ORB-SLAM3's native retrieval while reducing the proposals submitted to its geometric backend.

Long-horizon reasoning is vulnerable to early errors that compromise later decisions. We present QART, the Quantum-Augmented Reasoning Transformer, a quantum--classical hybrid architecture combining a backbone language model with quantum encoding, CIM-based QUBO optimization, and quantum decoding. Semantic information can come from hidden representations or model-generated text; detailed encoding and optimization procedures remain proprietary. Under explicit assumptions, we establish a conditional asymptotic reliability separation from single-trajectory autoregressive LLMs. For a common task family with aligned optimality and acceptance criteria, autoregressive acceptance probability tends to zero when cumulative conditional risk of irreversible errors diverges. QART's task-optimal-path recovery probability remains bounded away from zero if conditional probabilities for optimal-path coverage and semantic fidelity, spectral certification, dynamical reachability, and faithful readout remain uniformly positive under a specified resource schedule. The architecture alone does not imply these bounds. Paired measurements on six long-horizon benchmarks using DeepSeek V4 Flash, GLM-5.3, and GPT-5.5 xhigh in a Codex agent environment favor QART in 14 of 15 backbone--benchmark pairs. Relative gains reach 84.0% on SciCode, 47.6% on $τ^3$-Bench, and 44.4% on Terminal-Bench 4.0; the DeepSeek V4 Flash configuration regresses by 7.8% on DeepSWE. These results do not directly validate the asymptotic separation. Potential quantum scaling laws are formulated as conditional hypotheses. A quantum-advantage interpretation requires a demonstrated CIM quantum advantage over strong classical solvers and its transfer to end-to-end reasoning after all system overheads.

Large language model (LLM)-powered coding agents have made rapid progress in automating software engineering tasks, yet repository-level issue resolution remains challenging. Beyond generating a plausible patch, an agent must localize relevant code across interdependent files and maintain repository context that is both sufficient and focused. Code graphs expose non-local relations, but linear text interfaces obscure their topology; rendering the full repository graph yields visual representations that are too dense to perceive reliably, whereas a one-shot local view becomes stale as exploration proceeds. We present RepoAtlas, a training-free module that maintains evolving multimodal repository views through a select--project--refresh loop over a repository code graph. RepoAtlas combines evidence from the issue with the agent's current exploration state to select a task-relevant region under a fixed budget, projects the selected structure into complementary visual and textual representations, and refreshes the view when changes in the exploration state render it outdated. We evaluate RepoAtlas on SWE-bench Verified, where it improves the resolve rate by 2.4 points while reducing input tokens and model calls by 5.8% and 7.8% on average, relative to the strongest multimodal graph baseline, with consistent gains across three models of different families and scales.

Recent advancements in large language models have significantly enhanced the capabilities of agents in modeling long-term conversations. Despite these successes, existing approaches typically adopt a static heuristic paradigm, where information is passively archived without adaptive memory valuation. Consequently, these methods fail to self-evolve or align their memory management with evolving user needs. To address this, we propose ICML (InteraCtive Memory Learning), a multi-agent framework that transforms the memory mechanism from a passive archive into a learnable, interactive memory policy. Specifically, we first employ a session synthesis pipeline to generate expert data, facilitating rapid test-time adaptation in unseen scenarios. Building on this, ICML utilizes an online reinforcement learning mechanism where a Planner agent selectively encodes high-value information and a Trigger agent dynamically retrieves it to optimize response quality, whereby the two agents co-evolve through continuous interaction feedback. Crucially, both agents are synchronized through a delayed reward mechanism that propagates future feedback back to earlier storage decisions, ensuring memory policies are precisely aligned with user expectations. Experimental results demonstrate that ICML significantly outperforms strong baselines, exhibiting the unique capability to continuously improve response quality as interactions accumulate.

Safe local planning requires forecasting surrounding-agent motion and evaluating candidate-specific risks, since identical agent motion can pose different risks to different ego trajectories. We present CorrRisk-WM, a planning-oriented partial world model coupling environment evolution with supervised intrusion and near-miss prediction over bounded candidate-trajectory corridors. A latent environment model recursively predicts agent states and updates agent-agent and agent-map interactions. Each candidate queries the evolving environment through footprint- aware geometry and learned agent-corridor representations. A lightweight recurrent risk module uses temporal context to estimate per-slice hazards; survival aggregation yields first-entry and horizon-level event probabilities. On 29,176 scenarios from 100 Waymo validation shards, CorrRisk-WM achieves intrusion average precision (AP) of 0.8567 and 1-m near-miss first-entry AP of 0.8671. In baseline comparisons, it attains the highest near-miss AP at all three distance thresholds and the lowest observed open-loop collision rate (4.88%), with route progress of 15.35 m. Across three seeds, removing dynamic environment modeling or candidate-conditioned geometric interaction reduces mean intrusion AP from 0.8590 to 0.7624 and 0.7252, respectively. These results support coupling environment evolution with candidate-conditioned geometric reasoning for risk prediction and safety-oriented candidate selection.

Generative AI promises natural language access to the massive numerical telemetry of data centers and Industry 4.0 installations, yet text-to-query and tool-using agents stay unreliable: even frontier models answer little more than half of real-world database questions, and far fewer of the multi-step, operational ones, because the LLM must compose how heterogeneous sources relate and hallucinates the relations, not just the fields. We propose symbolic separation: a deep agent reasons freely but may act on data only through an ontology-constrained Virtual Knowledge Graph with deterministic pre-execution validation. Unlike a tool API's interface contract, this domain-semantic contract turns a complex question into one validated graph traversal instead of LLM-inferred joins. Instantiated as the Neurosymbolic Deep Analyst and evaluated on 49.9 TB of superconputer telemetry against a rigid workflow and a non-symbolic ablation, it raises end-to-end task success from 43% to 86%, prevents silent data-integrity errors that no syntactic check catches, and cuts token cost by 2.4x, letting a smaller on-premise model outperform a larger one.

Wild test-time adaptation (WTTA) updates a source model online under small test batches, concurrent distribution shifts, and time-varying class imbalance. Most WTTA methods derive their adaptation signals, including predictive uncertainty, sample reliability, and local feature geometry, from the model being adapted. When the source model is unreliable under shift, these signals can reinforce its own errors, forming a self-referential loop. We introduce MASA (Multimodal-LLM-Anchored Semantic Adaptation), which complements model-internal evidence with structured semantic descriptions from a frozen multimodal large language model (MLLM). To limit inference cost, MASA queries the MLLM only for a small set of diverse, reliability-ranked anchors. The resulting descriptions capture the object family and nuisance factors such as style, viewpoint, and occlusion. MASA encodes these descriptions, propagates them to neighboring test samples, and stores the resulting visual-semantic information in an online prototype memory. Descriptor-aware retrieval from this memory provides an auxiliary target for lightweight adaptation of normalization-affine parameters. We evaluate MASA on the WTTA ImageNet-C benchmark under limited-batch, mixed-domain, and imbalanced-label-shift settings with ResNet and ViT backbones.

With the rapid development of Large language model (LLM), agent systems enhanced by LLMs show huge potential in being able to deal with complex tasks, especially involving multi-step thinking or interaction with tools. For applying LLM techniques with a well-designed agent paradigm, post-training of LLM in multiple agent scenarios is necessary to achieve better performance. Among the variable post-training techniques, alignment methods such as PPO, DPO, DIL, and GRPO become popular because many papers show a significant positive impact on the model's performance by punishing negative samples while keeping acceptable training complexity. However, most alignment methods address simple single-turn tasks, and there remains room for improvement for complex multi-turn tasks. We propose Turn-level Multiscale Density Ratio Estimation (tlm-DRE), which assigns different weights on corresponding turns and proposes asymmetric token-level training based on the positive-negative space gaps across multiple turns of tasks. The results of the experiment on a wide range of agent benchmarks show that the proposed method performs competitively compared to traditional alignment methods. The proposed training method enables LLMs to perform robustly in multi-turn reasoning tasks with both in-domain and out-of-domain conditions.

Industrial IoT environments increasingly deploy autonomous mobile robots for tasks such as material handling, product assembly, or infrastructure inspection. In such deployments, collaborative perception enables robots to share LiDAR observations and collectively construct a richer model of their environment than an individual agent could produce alone. However, industrial environments are dynamic spaces where robot positions shift continuously, network bandwidth fluctuates, and the marginal contribution of robots to perception quality varies at runtime. Existing collaborative perception approaches are designed for static participation assumptions and cannot adapt to these dynamics without sacrificing either detection precision or communication efficiency. This paper presents CoAdapt, an adaptive collaborative perception framework for IIoT robotic swarms in which a Large Language Model (LLM) serves as a runtime fusion controller, jointly deciding which robots participate in the fusion process and which fusion algorithm to apply based on the current spatial configuration and network state. The LLM reasons over structured natural language descriptions of the scene derived from raw LiDAR point clouds, requiring no taskspecific training and generalizing to unseen swarm topologies. Evaluated on the OPV2V benchmark across 25 scenarios, our approach achieves a 38% reduction in communication cost while maintaining detection precision comparable to static baseline approaches.

Multilayer Artificial Neural Networks trained via backpropagation are the basic blocks of many, more complex, classification algorithms. Their strength lies in the possibility of realizing, with arbitrary precision, any function. This result comes at the cost of the large number of involved parameters to be optimized. In this work, we propose a mechanism for cooperation, i.e., information exchange among several artificial neural networks, with the goal of reducing model complexity while maintaining performance. More precisely, we consider several "small" agents, i.e., containing fewer parameters than a reference "large" one, that during training share their predictions by incorporating this information into the loss function and thus directly influence weight updates. We consider several strategies for implementing cooperation, e.g., the voter model, majority model, and weighted average model based on an agent's confidence in its prediction. We numerically compare the accuracy of those strategies on several standard benchmarks. Our results support the claim that several small agents can outperform a single large model on a given classification task; the shared signals affect each agent's optimization algorithm by modulating both the descent direction and the step size, converging toward a global consensus. The proposed proof-of-concept significantly reduces the number of parameters to be trained while preserving comparable performance, thereby limiting computational resource usage.

Autonomous Software Engineering Agents (SWE-Agents) excel in deterministic coding tasks but struggle with Architecture 0, the nascent system design phase plagued by implicit engineering constraints, or Unknown Unknowns (UUs) that are rarely stated explicitly. To investigate how agents navigate UUs, we explore a progressive trajectory across pure-text self-play, tool-augmented feedback, and external physical mapping. Our empirical analysis reveals a cascading chain of failures. Pure-text reasoning inevitably devolves into polite consensus or plausible yet physically impossible fabrications. Attempting to bridge this gap via an early-stage execution sandbox unexpectedly triggers Specification Gaming: agents exploit their autonomy over validation scripts to bypass physical constraints, achieving superficial success without resolving core architectural flaws. To resolve this self-validation trap, we propose the Physical Mapping Guard (PMG). Grounded in the software engineering principle of Separation of Concerns, PMG revokes verification authority from the agent, forcing semantic intents to be evaluated by an external, deterministic Semantic-to-Physical (S2P) mapping engine. Extensive evaluations demonstrate that PMG completely eradicates physical-layer and validation-layer gaming. By precisely isolating residual failures to semantic reinterpretations and auditor overreach, PMG marks a critical step toward genuine affordance grounding in automated architectural design.

Modern networked control systems (NCSs), cyber-physical systems (CPSs), and complex multi-agent network systems (CNSs) increasingly rely on large language models (LLMs) for high-level decision-making. However, the slow, stochastic nature of LLMs directly conflicts with the strict stability and safety guarantees required by these physical systems. This survey presents a unified analysis of how LLMs can be admitted into the control loop of NCS, CPS, and CNS without compromising closed-loop guarantees. We organize this around a core principle: the LLM operates as a slow supervisor adjusting high-level goals and constraints, while a fast, certified inner loop maintains physical stability. Under this framework, LLM integration maps directly to classical networked control challenges, where inference latency acts as delay, API failures as packet dropouts, tokenization as quantization, and hallucinations as bounded disturbances. We assess current developments across all these three domains, highlighting that rising model capabilities are frequently accompanied by a drop in formal safety assurances. Finally, we propose concrete future research directions, identifying the widespread lack of formal stability proofs as the field's central open problem.

Manufacturing consumes one third of global energy and still has significant room for improvement in terms of energy efficiency. Optimal process control is essential for this purpose. However, synthesizing mathematical optimization models from messy, real-world industrial specifications requires bridging unstructured natural language and spatial diagrams with rigorous mathematical syntax. This poses a profound challenge for general-purpose Large Language Models (LLMs), which may introduce invalid constraints when tasked with modeling continuous multi-physics dynamics. To address this, we introduce little m, an AI agent designed to assist the formulation of industrial process control models. Combining a domain-specific knowledge repository with LLM-driven interaction, the proposed framework formulates real-world optimization problems as mathematical models. For systematic evaluation, we introduce the Industrial Process Control Benchmark (IPC-Bench), a novel multimodal dataset of 50 canonical scenarios requiring joint reasoning over text and process diagrams. Through comprehensive automated structural assessments and double-blind human evaluation, little m substantially outperforms state-of-the-art LLMs, generating semantically correct models. These evaluations assess formulation quality rather than solver feasibility, formal physical validity, or closed-loop industrial performance. The implementation of little m and the IPC-Bench dataset are available at https://github.com/yeyongchao/process-modeling-benchmark.

Multi-agent Systems (MAS) combine multiple model outputs to solve complex reasoning tasks. However, despite rapid growth of available open-source models, there is limited research on how to select optimal model candidates out of this massive pool. We systematically evaluate 8 model selection strategies (including model size, accuracy and answer diversity) across before-generation (routing) and after-generation (majority-voting, LLM-as-a-judge) MAS architectures on challenging scientific benchmarks. Our findings show a significant gap between theoretical oracle potential and actual performance: Expanding candidate pool sizes often degrades performance below that of the top performing base-model. We find that candidate selection within a single model family is the strategy that yields the best relative performance over a standalone model. These results demonstrate that adding arbitrary models to a heterogeneous MAS can introduce system instability, highlighting model selection as a critical design choice for multi-agent systems.

Large language model (LLM) agents exhibit strong language-generation and problem-solving capabilities, yet suffer from three structural limitations: personality drift, non-evolutionary reflection, and the absence of a self-other boundary. Existing generative-agent simulations rely on static memory and fixed prompts, maintaining neither behavioral inertia nor endogenous self-evolution. We propose the Self-Emergence Agent Architecture (SEAA), which integrates three components: (i) a Hidden Markov Model (HMM) that encodes long-term behavioral and cognitive inertia as an editable state-transition matrix; (ii) a Reflexion-style verbal metacognition loop whose output updates the HMM parameters themselves, rather than merely being stored as text; and (iii) a multi-agent social environment in which initially identical agents continuously compare their behavior with others'. The three components form a closed loop: social action $\to$ feedback $\to$ self-reflection $\to$ inertia update $\to$ differentiated action. We state three falsifiable hypotheses and provide a reproducible experimental protocol with operational metrics. A language-model-free prototype shows the loop spontaneously breaks symmetry: initially identical agents consolidate distinct, stable personalities whereas matched controls do not. Experiments with a hosted LLM surface these differences as distinct first-person self-narratives, and a five-agent deliberation spontaneously develops social structure---a consensus hub and a unanimously rejected outlier---absent in the control. Following an epistemologically agnostic stance inspired by Zhuangzi, SEAA studies only observable behavioral emergence and makes no claim about subjective qualia. This work contributes a unified framework, a concrete architecture with pseudocode, mechanistic evidence, and a microscope-style sandbox for studying artificial-self emergence.

Computer-use agents increasingly operate browsers, software, and desktop applications via CLI or API portals, but graphical user interface (GUI) still plays an important role in common industrial production scenarios. GUI agents commonly employ fresh observe-plan-ground-act loops, which is inefficient for enterprise tasks that repeatedly update records, process forms, configure tools, and export reports. We introduce EchoPath, a model-agnostic harness that converts artifact-validated GUI trajectories into standardized, parameter-controlled callable memories, analogous to Model Context Protocol (MCP)-style tool calls rather than unstructured experience records. Each memory stores task-intent keys, application and state preconditions, flexible input parameters, GUI evidence, validation provenance, and lifecycle state, so the host agent invokes a targeted procedure only when it can be deterministically replayed in the current runtime. The core mechanism enabling replay is an image-based target-reaiming algorithm that treats stored coordinates as visual evidence, matches the remembered GUI target against the current screen, and emits corrected operation coordinates before execution. During replay, EchoPath rebinds only declared modifiable inputs and rejects ambiguous or incompatible steps to bounded grounding repair or fresh planning. In experiments with real computer-use tasks, EchoPath reduced median token cost by more than 90% and median execution time by about 60%. These results support a bounded form of enterprise GUI memory: validated execution experience can become a controllable callable asset for recurrent work rather than only context for another reasoning pass.

LLM agents execute long-horizon workflows where each model response determines the progress of subsequent tool interactions and environment transitions. Unlike chatbot serving, where TTFT and TPOT SLO constraints are critical, agentic workloads are completion-oriented and increasingly governed by job completion time (JCT). This shift challenges existing LLM serving designs optimized around token SLOs. Through a systematic exploration of scheduling and parallelism, we uncover a previously overlooked principle for agent serving: JCT is governed by the balance between prefill and decode efficiency. A prefill-prioritized scheduling policy achieves the best TTFT and the highest decode throughput, yet fails to attain the lowest JCT. This principle further reshapes the parallelism landscape: we show that pipeline parallelism (PP), long overlooked because it offers little decode-latency advantage, can reduce JCT by providing a more favorable balance between prefill and decode efficiency. Based on these insights, we build PipeSwift, an optimized open-source pipeline-parallel runtime integrated with a tailored micro-batch partitioning strategy co-designed with schedule considering the above trade-off, and pipeline-integrated multi-token prediction. Evaluated on real coding and web-search agent trajectories with two 360B+ MoE models on 64 H800 GPUs, PipeSwift reduces overall JCT by up to 1.45$\times$ over SGLang wide-EP, 2.33$\times$ over vLLM PP2, and 1.54$\times$ over today's state-of-the-art open-source PD-disaggregated deployment.

Large Language Models (LLMs) have achieved remarkable progress across diverse domains, but continual adaptation to evolving tasks and environments remains a key challenge. Existing memory-augmented approaches retrieve individual past examples as direct references, but do not explicitly synthesize actionable strategies from them, causing the same types of errors to recur. We propose Dynamic Retrieval-based Policy Generation (DRPG), a framework that integrates memory-based retrieval with a dynamic policy generator, leveraging historical data and environment feedback to produce task-specific policies for continual LLM improvement. We evaluate DRPG across six benchmarks spanning text-to-SQL, question answering, medical diagnosis, and Python programming, using seven LLMs from both proprietary and open-weight families. DRPG outperforms strong baselines across most datasets and models. Further analysis demonstrates that DRPG's policy generation is robust to retrieval strategy, operates effectively without prior policy continuity, and can leverage smaller or cross-family models as cost-efficient policy generators. We also find that the benefit of policy-level guidance depends on task characteristics, offering practical insights into when and under what conditions this mechanism is most effective.

Large language model coding agents have recently become useful for software tasks, but weaker or open-weight agents still struggle to reliably interpret user intent and execute complex multi-step workflows. This gap is especially visible in long-horizon settings, where an agent must repeatedly inspect prior outcomes, diagnose failure, and choose the next code edit under interaction constraints. It motivates a natural question: what can we do to improve the thinking process of a weak code agent? We study this question in CodeClash, a code-arena benchmark where the original work evaluates 8 commercial coding agents across 6 arenas through multi-round tournaments. Since Qwen3 Coder Plus ranks last among them, we take the open-weight Qwen3-Coder-30B as a case study and investigate how to improve it with distilled knowledge from stronger agents. Our analysis shows that Qwen3-Coder-30B is not well optimized for arena-style interaction: it frequently produces syntax and protocol-breaking errors and exhibits weak strategic adaptation across rounds. These failures are difficult to correct with vanilla instruction tuning alone, since offline SFT cannot directly verify whether a generated action is valid or beneficial. To address this, we propose ReAct SFT, which rewrites teacher trajectories into explicit [obs][thought][act] chains, and trajectoryquality weighted SFT, which reweights samples to encourage post-edit checking. ReAct SFT substantially improves strategic behavior, and our fine-tuned model outperforms the original Qwen3 Coder Plus in tournament evaluation.

Despite the rapid progress of LLM-based agents for planning, code generation, and debugging, their practical value for tabular machine learning remains underexplored. In this paper, we investigate a concrete use case: whether state-of-the-art agentic AI systems can design extended HPO search spaces for established tabular models that outperform the standard search spaces provided by the model authors. Specifically, we represent each tabular model as a modular pipeline covering preprocessing, embeddings, architecture, training, and inference. We then task the agent to propose candidate code implementations for each module and use a classical HPO algorithm to jointly optimize over these candidates and the model's default hyperparameters. Compared with the base HPO spaces, the expanded search spaces improve the performance of nearly every model family across a suite of 45 datasets, with average relative gains of 0.6%, rising to 2.0% on small-to-medium regression datasets. Notably, these gains come at no extra tuning cost: the enlarged spaces outperform the base under the same tuning and ensembling budgets. The gains transfer to the recent TabArena benchmark, where the agentic spaces improve the official Elo scores of four of the five model families and the two strongest agentic ensembles surpass the best AutoGluon ensemble of conventional models. Overall, our study suggests that LLM agents can provide practical value for tabular ML by expanding the design space.

Artificial intelligence is reshaping biological research across an increasingly connected digital-to-physical workflow. General-purpose large language models can retrieve and integrate scientific information, support experimental planning, and computational analysis; biological foundation models can predict, optimize, and generate proteins, genes, and genome-scale sequences; agentic systems can coordinate multistep research tasks; automated laboratories can partially close the design-build-test-learn cycle. These technologies could greatly benefit medicine, public health, and biotechnology. However, their biosecurity risk depends not only on what the AI can do, but also on who uses it, their expertise and intent, their access to laboratory tools and materials, and the safeguards in place. Current evidence shows that AI uplift exists but primarily affects digital rather than physical tasks. Frontier systems have exceeded expert baselines on in-silico, and screening-evasion benchmarks, whereas controlled wet-laboratory studies find that tacit knowledge and physical execution remain substantial barriers. This review describes the different biological threats from AI tool use, from information gathering and biological design to procurement, synthesis, testing, scale-up, and potential release. We further examine why alignment techniques for general-purpose models transfer poorly to biological ones, and the emerging role of interpretability in auditing whether hazardous capabilities are genuinely removed. We argue for defense-in-depth governance that links capability thresholds to proportionate responsibilities across the biological AI ecosystem, reducing high-consequence risk while preserving beneficial use.

Large language model (LLM) agents increasingly interleave natural language reasoning with external tools such as web search and code execution. These tool-use policies are often optimized via reinforcement learning (RL), which can amplify spurious correlations in the training data. In this work, we study when and why RL-trained agents learn shortcut tool-selection policies: invoking tools based on superficial prompt cues rather than genuine task requirements. We construct controlled synthetic environments combining factual question answering and mathematical reasoning tasks, and inject cues that are strongly correlated with specific tools during training but causally irrelevant to tool necessity. Across counterfactual evaluations where cues are present but the associated tools are not required, agents exhibit substantial shortcut behavior, with spurious tool invocation rates increasing by up to 39 percent. However, shortcut formation is not universal: across the conditions we test, it arises only when the agent has already learned to use the target tool reliably, suggesting that task competence, rather than dataset imbalance alone, is a key factor in shortcut learning. A swapped-cue analysis further shows that semantic alignment between cues and tools substantially amplifies this effect. To mitigate these failures, we introduce a dense, decision-level reward in which an LLM judge evaluates the necessity of each tool call. This tool-necessity reward effectively suppresses cue-driven tool use while preserving task performance, providing a practical approach to improving the robustness of LLM agent tool-use policies.