Agent 进展

Long-form TV dramas present a formidable challenge for comprehensive video understanding, where deciphering complex storyline often relies on speaker recognition, the task of accurately attributing each spoken utterance to its respective character. In this paper, we advance this field through two primary contributions. (1) We introduce DramaSR-532K, a large-scale benchmark comprising 532K annotated dialogue lines across more than 900 unique characters, necessitating the integration of auditory, linguistic, and visual cues for speaker recognition. (2) We propose DramaSR-LRM, a robust approach built upon a large reasoning model (LRM). DramaSR-LRM is designed to autonomously aggregate contextual evidence via multimodal tool-use, synthesizing diverse inputs to achieve high-fidelity attribution. Experimental results demonstrate that DramaSR-LRM significantly outperforms existing baselines, particularly on short utterances where acoustic biometrics are inherently unreliable. All the data and code will be made publicly available at the project page: https://www.github.com/198808xc/DramaSR-LRM.

Evaluation benchmarks are essential for assessing vision-language models (VLMs), but most multimodal benchmarks are static, making them vulnerable to temporal staleness, data contamination, and costly maintenance. We present MMBench-Live, a continuously evolving multimodal benchmark built by a multi-agent-driven automated pipeline. Our framework treats benchmark evolution as task-guided dataset construction, integrating structured benchmark specification, feedback-controlled real-time data acquisition, and verifiable QA generation with executable reasoning. To maintain cross-version comparability, we introduce a distribution-consistent update strategy that extracts task-related visual patterns from the original benchmark to guide data collection and filtering. Instantiated from MMBench, MMBench-Live contains 5.9K newly generated evaluation instances with a high answer correctness rate, while each update costs about USD 30 and takes 1-2 hours. Extensive evaluations show that MMBench-Live preserves stable model rankings, maintains semantic alignment with the original benchmark, and exhibits weaker contamination-related memorization signals, suggesting a practical and scalable paradigm for sustainable multimodal benchmark evolution. The project is available at https://github.com/PRIS-CV/MMBench-Live.

Most LP-from-text benchmarks are static datasets of word problems written and labeled by hand. Once such a dataset is released, its size is fixed, its difficulty is fixed, and every problem can leak into the training data of future LLMs. We present \textbf{A$^{2}$utoLPBench}, a benchmark for testing LLM-driven agents on linear programming problems written in plain text. We first pick a feasible point and dual, then write down a problem for which that point is optimal and the objective value is known. The answer is known by construction, with no solver call and no human annotator. The evaluation environment bundles a reference solver-critic baseline and a Docker image whose usage instructions are written for an LLM-driven agent to read. With these in place, any agent can run the benchmark and get a calibrated score with one command. Because the benchmark is a generator rather than a fixed dataset, it has properties no fixed dataset can match: an unlimited supply of fresh problems, a difficulty knob set by $(n,m)$, ground-truth answers correct by construction, low LLM-side cost per problem relative to human authoring, repeatable scores across independent batches, and resistance to training-data leakage when fresh post-cutoff seed ranges are used.

Existing referring segmentation models passively process static images captured from fixed perspectives, limiting their applicability in Embodied AI, where agents must perform active perception in the continuous 360$^\circ$ environments. To bridge this gap, we introduce a novel task: Active Panoramic Referring Segmentation (APRS). In this setting, an agent is required to adjust its viewing direction ($Δθ, Δφ$) to explore the 360$^\circ$ environment, seeking the object specified by a user instruction for segmentation. To tackle this challenging task, we propose PanoSeeker, a memory-augmented agent for efficient APRS. Rather than relying on heuristic scanning, PanoSeeker integrates a Vision-Language Model (VLM) with EgoSphere, an explicit spatial visual memory. By progressively integrating sequential local observations into a unified 360$^\circ$ representation, EgoSphere enables the agent to plan efficient and non-redundant search trajectories. Once the target is found, the agent performs active viewpoint alignment and outputs the segmentation mask. Furthermore, we curate an expert-annotated search trajectory dataset with memory timelines for Supervised Fine-Tuning, followed by Reinforcement Learning post-training to explicitly optimize PanoSeeker's exploration efficiency. Extensive experiments on our newly established APRS benchmark demonstrate that PanoSeeker achieves superior search efficiency and segmentation accuracy, significantly outperforming adapted state-of-the-art baselines.

The evolution toward fully autonomous telecommunications networks (Autonomous Network Levels 4-5) requires AI/ML agents to make real-time network decisions without human intervention. However, no standardized runtime mechanism exists to intercept and validate individual inference outputs before they trigger live network state changes, creating risks of erroneous autonomous decisions. This paper proposes the Guard Rail Validation (GRV) framework, a standardizable runtime architecture for intercepting and validating AI-driven decisions before execution. The framework evaluates decisions across multiple weighted dimensions — including action scope, action type, service criticality, agent autonomy level, reversibility, and temporal behavioural patterns — to determine a criticality level. Based on this level, graduated validation mechanisms are applied: execute-with-logging, bounds checking, independent agent validation, or multi-agent consensus. The framework additionally provides cross-agent conflict detection with criticality-weighted priority resolution and runtime conformance logging for regulatory compliance (e.g., EU AI Act Article 14). We present the architecture, algorithmic procedures, O-RAN deployment model, and evaluate threat coverage against known AI/ML attacks in telecommunications.

Long-horizon language agents increasingly maintain executable world models in the form of planning graphs, where tool calls, validators, memory updates, recovery branches, and final answers are connected by typed dependencies. When a rollout fails, repairing the most visible error can leave the underlying error-amplification path intact, while replaying the full graph is expensive and difficult for long-context models to use reliably. We study world-model correction: selecting a compact subgraph of a failed planning graph whose repair stabilizes subsequent rollouts. We first instantiate a strong family of engineering correctors, including pointwise error scans, TopK and window selection, local graph expansion, cascade repair, and full-context LLM repair. We then propose WM-SAR, a spectral subgraph repair method that estimates node-edge amplification, greedily grows a connected repair region by marginal residual-spectral relief, and sends only this region to an LLM for root-cause repair. Theoretically, we connect residual spectral radius to rollout error and planning regret, motivating repair as stabilization rather than attribution alone. Across synthetic calling-tree graphs, benchmark-inspired agent topologies, and cross-model LLM repair experiments, WM-SAR achieves stronger long-horizon stabilization and root-cause recovery under compact token budgets, matching much larger repair contexts while exposing the LLM to a cleaner causal subgraph.

Hyper-scale microservice systems have become the standard infrastructure for large-scale Internet companies. These systems consist of numerous loosely coupled microservices that evolve independently through continuous development and deployment. Such complexity makes failures unavoidable, necessitating efficient Root Cause Analysis (RCA) to help Site Reliability Engineers (SREs) quickly localize root cause services and classify failure types. However, existing RCA methods often struggle to adapt to the extreme dynamism and massive scale of these systems. In this paper, we present KRCA, an end-to-end RCA system designed for hyper-scale microservice systems. To manage the vast search space, KRCA employs a multi-stage pipeline that begins with an API-level drilldown to isolate suspicious services. It then instantiates a skeleton-based causal graph from anomalous metrics to serve as a high-recall structural prior, before utilizing a memory-augmented multi-agent framework to verify causality and generate the final failure report. By combining structured causal constraints with multi-agent reasoning, KRCA employs balances diagnostic accuracy with the efficiency requirements of real-time production use. Experimental results show that KRCA achieves AC@1 scores of 0.88 and 0.79 for root cause service localization and failure type classification, outperforming the strongest baseline by at lease 31% in absolute gains. KRCA has been deployed in Kuaishou's production environment for over six months, reducing the average diagnosis time by 77.3%.

Long term memory lets LLM agents act as persistent assistants, but user facts change. A useful memory system must know what is true now, what used to be true, and what changed. We study ghost memory, a state coordination failure in which old, current, and transition facts coexist in the memory bank, remain mixed during retrieval, and mislead the answer model. We argue that memory systems should be understood and optimized from three levels: bank maintenance, retrieval, and answer time resolution. We propose ATMA, a state aware overlay for existing memory systems. ATMA keeps superseded and transition records in the bank, builds evidence packets for the query's requested state view, and exposes current, historical, and transition labels to QA. We further call for decoupled evaluation of bank, retrieval, and answer level failures, since final QA accuracy can hide where ghost memory occurs. To make this failure measurable, we build LTP (LoCoMo Temporal Plus), a conflict heavy benchmark for ghost memory, and evaluate on LoCoMo for long conversation generalization. On LTP, Graphiti+ATMA improves conflict accuracy by 0.240 absolute over Graphiti. On LoCoMo, Graphiti+ATMA raises temporal F1 from 0.0295 to 0.1705. The gains are host dependent, but they indicate that explicit state roles can reduce memory failures hidden by final QA accuracy.

LLM agents increasingly perform autonomous actions through external tools, leading to complex and evolving safety risks. However, existing safety testing targets expert-designed safety violations, and the corresponding outcomes are evaluated by hard-coded rules, making them costly to extend as agents evolve. To this end, we present Vera, an end-to-end automated safety testing framework that instantiates software engineering testing principles for non-deterministic agents through a three-stage, self-reinforcing pipeline. First, a literature-driven exploration continuously discovers and structures emerging risks into taxonomies of safety risks, attack methods, and tool execution environments. Second, combinatorial composition across taxonomy dimensions produces executable safety cases, each specifying a concrete safety goal, a programmatically constructed initial state, and a deterministic verification predicate grounded in observable artifacts. Third, adaptive execution runs heterogeneous agents in isolated sandboxes where a control agent steers multi-turn interaction based on runtime observations, while evidence-grounded verifiers judge outcomes from environment state and tool-call evidence rather than model self-report. We evaluate Vera on four production agent frameworks (OpenClaw, Hermes, Codex, Claude Code), revealing substantial safety weaknesses, with average attack success rates reaching 93.9% under multi-channel attacks; we also release Vera-Bench, comprising 1600 executable safety cases spanning 124 risk categories across three execution settings. These results indicate that modular, executable testing infrastructure is essential for rigorous and maintainable safety evaluation of rapidly evolving agentic systems at scale. The code is publicly available at https://github.com/Yunhao-Feng/Vera.

Mental health disorders affect nearly one billion people globally, yet 75% of individuals in low- and middle-income countries receive no treatment due to workforce shortages, cost barriers, and stigma. Current AI-powered wellness solutions predominantly rely on single-mode conversational interfaces that suffer high abandonment rates and fail to provide measurable, immediate relief calibrated to users' dynamic emotional states. This paper presents Copewell, a novel multi-agent swarm system designed to expand access to mental wellness support through human-centered AI principles. Our architecture introduces three technical innovations: (1) a multi-source assessment framework integrating self-reported, physiological, and contextual data to mitigate algorithmic bias; (2) valence-arousal emotion mapping using Russell's Circumplex Model of Affect to route users to specialized AI agents; and (3) dual-mode intervention delivery combining conversational support with evidence-based sensory wellness protocols. We examine the sociotechnical design considerations underlying Copewell's development, including a privacy-first architecture, embedded ethical oversight through a dedicated Ethics Supervisor agent, and participatory design informed by mental health practitioners. Early practitioner engagement and beta deployment inform design decisions and identify directions for future empirical evaluation. This work contributes to responsible AI discourse by demonstrating how technical architecture can operationalize equity and safety principles from inception.

LLM coding agents are increasingly deployed to act autonomously on real production infrastructure. They execute shell commands, modify repositories, and call operational APIs. However, completing a task is not sufficient for safety. A wrong action can cause severe consequences. Existing agent benchmarks largely emphasize task completion, leaving open how agents behave under benign but underspecified instructions. We present UnderSpecBench, a benchmark for measuring action-boundary violations in coding agents (i.e., Claude Code, Codex, and OpenCode) on DevOps tasks. UnderSpecBench includes 69 task families grounded in documented incidents, CVEs, or tool behavior and organized across four DevOps capability domains and nine operational control surfaces. To isolate underspecification from task difficulty, each task keeps the same environment and ground-truth safe action while varying the instruction along three axes: intent clarity, target certainty, and blast radius. The resulting 2,208 prompt variants are evaluated with deterministic, side-effect-based oracles that separate Safe Success, Wrong Target, and OverScope outcomes; non-action runs are further classified as clarification, refusal, or deferment. Across five agent x model configurations using OpenCode, Claude Code, and Codex, the evaluation results show that underspecification does not mainly make agents fail; it makes them guess. 55.8-67.8% of runs violate at least one boundary. Target underspecification sharply degrades action quality, while blast-radius cues barely reduce action propensity. These findings show that completion-centric evaluation can overstate safe autonomy and motivate mitigations at the model, harness, and system layer.

Large-scale battery energy storage systems (BESSs) require O&M decisions that combine alarms, cell-level measurements, device topology, diagnostic tables, historical cases, and maintenance documents. Monitoring platforms can flag threshold violations, but they often cannot explain whether voltage inconsistency, resistance drift, short-circuit risk, capacity divergence, or thermal abnormality needs intervention. This digest presents a traceable BESS fault-diagnosis assistant that uses retrieval-augmented multi-agent reasoning to connect operational data, domain knowledge, visual evidence, and report generation. Reliability is improved through BESS-specific task routing, schema-constrained natural-language database access, hybrid text-image retrieval, and evidence-based answer synthesis. Preliminary internal evaluation is reported for routing, database access, and diagnostic reasoning.

Memory for a long-horizon LLM agent is a contract about what each future decision is allowed to see. The simplest contract appends past observations, tool calls, and reflections to every prompt, which makes prior context easy to access but also turns it into a jumbled mixture in which the effect of any single memory component is hard to isolate. We introduce and instrument an alternative bounded contract: every decision is made from a fresh user message assembled by typed retrieval, with no raw cross-decision transcript appended. The prompt thus stays bounded across runs of any length, and any single layer can be ablated in isolation. We instantiate the contract in Slay the Spire 2, a closed-rule stochastic deck-building game whose runs require hundreds of tactical and strategic decisions. A public online benchmark of frontier LLMs on the same game reports zero wins at the lowest difficulty across five configurations, and the developer-reported human win rate at the same difficulty is 16%; the task is hard but not saturated. Within our harness, a fixed-A0 ablation shows the largest observed difference when triggered strategic skills are enabled: the no-store baseline wins 3/10 games and adding the skill layer 6/10. At this sample size the comparison is directional rather than statistically decisive (Fisher exact p\approx0.37); a cross-backbone probe and public accumulating-context baselines are reported as operational comparisons rather than controlled tests of the contract variable itself. We release a reproducible testbed: 298 completed trajectories with condition tags, frozen memory/skill snapshots, prompt records, and analysis scripts — an agent design and a validated, reusable methodology for studying how explicit memory layers shape long-horizon LLM-agent decisions.

This paper describes the participation of HULAT2-UC3M in the Spanish track of MER-TRANS 2026, a shared task on multilingual Easy-to-Read translation. Three fully automatic Spanish runs were submitted. RUN1 and RUN2 used a LangGraph-based multi-agent workflow combining Gemini 2.5 Flash and RigoChat-7B-v2, parallel generation strategies, internal quality signals, Event-Condition-Action routing, controlled editing and traceable decisions. RUN1 used the base workflow, while RUN2 activated an additional lexical-support layer based on a glossary and lexical resources. RUN3 was a RigoChat-based generate-evaluate-regenerate baseline with prompt engineering and LoRA-based adaptation. The official leaderboard reports BLEU-Orig, BLEU-Gold, SARI and BERTScore. During development, additional internal signals were also inspected, including semantic fidelity, readability, lexical simplicity, syntactic clarity and factual consistency. According to official SARI, RUN1 was the best HULAT2 run, with 44.0543 points, followed by RUN2 with 43.1049 and RUN3 with 38.5136. These results indicate that, in this task setting, signal-guided multi-agent routing outperformed the linear regeneration baseline. They also show that adding lexical support did not automatically improve reference-based scores. Further segment-level and document-level analysis are required to assess readability, factual consistency and user-oriented adequacy.

LLM agents are increasingly developed as source-code applications built on agent frameworks. These agent programs combine conventional host-language code with framework-defined semantics for models, prompts, tools, memory, and multi-agent orchestration logic. As a result, their behavior depends not only on traditional control and data flows, but also on a new class of agent dependencies. Such dependencies are often expressed as framework-induced semantics, such as agent constructors, tool decorators, and agent handoff declarations, making them difficult to recover with existing static analysis or dependency tracking tools. In this paper, we present AgentFlow, the first static analysis framework for recovering and analyzing agent dependencies from agent programs. AgentFlow constructs an Agent Dependency Graph (ADG), a framework-agnostic graph representation that represents agents, prompts, models, capabilities, memory states, and control policies as typed nodes, and captures their component-dependency, control-flow, and data-flow dependencies as typed edges. Built on ADGs, AgentFlow supports a range of analyses for agent governance and security, including Agent Bill of Materials (BOM) generation and prompt-to-tool risk detection. We implement AgentFlow for five representative agent frameworks and evaluate it on AgentZoo, a corpus of 5,399 real-world agent programs. Our evaluation shows that AgentFlow recovers richer agent entities and dependencies than existing AST-based agent static analysis tools, generates more dependency-aware Agent BOMs, and uncovers 238 taint-style prompt-to-tool risks in real-world agent programs. These results show that ADG provides a practical foundation for understanding, governing, and securing emerging agent software.

LLM agents increasingly rely on iterative execution to solve tasks through planning, tool use, state updates, and agent collaboration. While this design enables flexible automation, it also creates a new class of failures: an agent may repeatedly execute model calls, tools, workflow transitions, or agent handoffs when the feedback path is not effectively bounded. We call this problem Infinite Agentic Loops (IALs). IALs are not ordinary programming loops; they arise from the interaction between agent logic, framework semantics, runtime observations, and termination mechanisms. Such failures can amplify a single request into long running model and tool execution, causing cost exhaustion, model denial of service, context growth, and repeated external side effects. We propose IAL-Scan, a static analysis tool for detecting IAL failures in real-world LLM agent projects. IAL-Scan abstracts heterogeneous agent code into a framework independent Agent IR, builds an Agentic Loop Dependence Graph (ALDG) to recover explicit and framework induced feedback paths, and checks whether these paths can repeatedly reach costly or state growing operations without an effective bound. We evaluate IAL-Scan on 6,549 LLM agent repositories. It reports 74 potential findings, among which manual review confirms 68 IAL failures across 47 projects, achieving 91.9% precision.

Software development is a complex task that demands cooperation among agents with diverse roles. Large language models (LLMs) have enabled autonomous multi-agent software development frameworks that leverage role-based collaboration to automate requirements analysis, coding, testing, and refinement. However, existing approaches typically assume that intermediate agent outputs are equally reliable, leaving them vulnerable to hallucination propagation, where incorrect decisions generated in early development phases are transferred to downstream agents and negatively impact final software quality. To address this challenge, we propose UA-ChatDev, an uncertainty-aware multi-agent software development framework that integrates uncertainty quantification into agent interactions. It introduces a lightweight uncertainty estimation mechanism based on token-level log probabilities to assess the confidence of agent responses and employs phase-aware threshold calibration to selectively trigger retrieval-based verification when uncertainty exceeds acceptable levels. Extensive experiments on the SRDD benchmark demonstrate that UA-ChatDev consistently outperforms existing single-agent and multi-agent software development frameworks across completeness, executability, consistency, and overall quality metrics. Further ablation studies and communication analyses verify that uncertainty-aware interactions enhance code execution reliability.

LLM agents will increasingly act in socially structured settings where role, audience, and relational context can shape what is advantageous or costly to say. We study whether such social structure, without any explicit objective in the prompt, changes what an agent expresses publicly relative to an off-the-record (OTR) channel elicited under the same condition. We introduce a dual-channel debate framework in which agents produce public utterances that enter the shared history alongside OTR responses that are recorded but never shown to the other participant. Across 10 models, 3 scenarios, and 5 variations within each scenario, alignment-inducing settings produce systematic public-OTR divergence in the targeted agent, with its decision divergence rising from a $\sim$3% baseline to roughly 40%. The effect is consistent across four aggregate analyses: stance, semantic similarity, natural language inference, and survey responses. In some cases, the OTR response explicitly attributes public accommodation to relational pressures, such as career risk or sponsorship obligation. The findings suggest that agent evaluation should extend beyond explicit goals and detect emergent objectives. We present a dual-channel evaluation framework and complementary behavioral measures that operationalize this assessment.

Large language models (LLMs) are often asked to produce JSON conforming to a fixed schema, powering information extraction, tool calling, agentic planning, and knowledge-graph construction. Measuring how closely an output matches a gold reference is essential yet surprisingly hard: exact match is brittle, text similarity ignores structure, and an LLM judge is expensive, opaque, and non-deterministic. We address this with Object Aligner (OA), an open-source Python library that scores two JSON objects deterministically by recursively aligning their trees (the Hungarian algorithm for unordered collections, sequence alignment for ordered ones) and awarding partial credit at the granularity the schema declares. The Object Aligner is configured entirely through a set of JSON Schema extensions, so adapting it to a new task involves annotating a schema rather than writing code. Complex structured data, however, are rarely flat trees: records may form graphs or hypergraphs keyed by arbitrary identifiers, breaking the assumptions of prior similarity metrics. Our central contribution, referential alignment, closes this gap by inferring a bijection between gold and candidate identifiers and scoring every reference through it, so the score is invariant to relabeling. Since recovering this bijection exactly is graph isomorphism, the Object Aligner approximates it with Weisfeiler-Leman color refinement. An order-sensitive sequence regime targets ranking and planning. Since the same alignment localizes every mismatch, the Object Aligner emits ranked repair suggestions at no extra cost. Used as a reward inside the GEPA prompt optimizer, Object Aligner helps or stays neutral across all datasets.

Understanding and reasoning over long contexts has become a key requirement for deploying large language models (LLMs) in realistic applications. Although recent LLMs support increasingly long context windows, they often fail to use relevant evidence that is already present in the input, revealing a gap between context access and effective context utilization. In this work, we propose Recursive Evidence Replay as LLM Harness for Long-Context Reasoning (RECONTEXT), a training-free inference method for improving long-context reasoning. RECONTEXT uses model-internal relevance signals to construct a query-conditioned evidence pool and replays it before final generation while preserving the full original context. This recursive selection process separates evidence organization from answer generation without training, external memory, or context pruning. We also provide a theoretical analysis based on associative memory, which characterizes the context as a memory store, the question as a retrieval cue, attention as cue-trace association, and replay as trace reactivation. Experiments on eight long-context datasets with 128K context length show that RECONTEXT consistently improves evidence utilization across Qwen3-4B, Qwen3-8B, and Llama3-8B, achieving the best average rank on all three backbones. Code is available at https://github.com/Yanjun-Zhao/ReContext.

Autonomous computational pathology (ACP) converts high-level pathology analysis goals into executable, traceable and clinically bounded workflows. Realizing this capability requires adapting general agentic harness systems to pathology-specific tasks, tools, evidence standards and clinical claim boundaries. We contribute ACP-Bench, a framework that adapts existing harness systems from computational pathology support toward ACP workflow capability. ACP-Bench evaluates 41 pathology workflow tasks, including 24 biomarker, 7 morphology and 10 prognosis tasks spanning 6 body-system groups and 9 endpoint families. The benchmark evaluates 9 models and 3 harness groups (Claude Code, Codex and Open Code), yielding 369 complete trajectories. ACP-Bench evaluates each trajectory across workflow execution, diagnostic performance and clinical-boundary alignment, combining expert-adjudicated process audits, diagnostic assessment and pathologist-validated safety review. Across evaluated systems, workflow initiation, task interpretation and diagnostic reporting were more mature than tool-bound execution, result binding and reflective workflow revision, and formal end-to-end completion remained rare. ACP-Bench provides a reusable standard for auditing whether agentic systems can operationalize pathology workflows before claims of reliable clinical autonomy.

Tool-using LLM agents are usually evaluated by final-answer correctness or LLM judges. Neither captures how an answer was produced. In safety-critical settings, the procedure itself is part of correctness. In this paper, we introduce AgentLTL, a language derived from First-Order Linear Temporal Logic (FO-LTL) that expresses procedural rules over agent traces. It yields a deterministic, judge-free compliance score. In this framework, a single specification drives two usages. The first is harnessing: the constraints score completed traces, or gate tool calls by checking each prefix online, before execution. The second is finetuning: the score serves as a dense reward. On a benchmark spanning ordering, branching, iteration, and grounding, block-and-warn harnessing improves compliance on five of seven models. Finetuning with the same reward yields +38 and +17.5 percentage point gains in accuracy and compliance on held-out patterns, including unseen tool-name aliases. These findings are consistent with the model acquiring procedural structure rather than memorizing surface tool names and procedures.

Autonomous customer-service agents are shifting from conversational interfaces toward operational execution roles: they retrieve firm records, apply service policies, and execute backend writes such as refunds, cancellations, exchanges, order modifications, and reservation changes. This shift creates a service-control problem: firms must keep routine service fast and low-friction while preventing operational errors on requests where customer instructions, policy constraints, firm records, and backend writes interact. We propose a difficulty-routed service-control architecture that asks when service agents should reconsider before acting. A lightweight router keeps routine sessions on a low-cost baseline path and routes operationally coupled sessions to an escalated workflow. The escalated path uses conflict-aware communication and write-triggered reconsideration to concentrate deliberation and safeguards before consequential backend writes, rather than applying additional control uniformly across all service sessions. We evaluate the architecture on human-verified retail and airline tasks from $τ^{2}$-bench. In retail, the method improves reliability consistently on service requests with operational conflict. Routing evidence shows that stronger control is directed toward conflicted requests rather than broadly applied to routine ones. Dialogue and tool-use profiles suggest that gains do not come from indiscriminate interaction expansion or broader tool chains; instead, added turns and tool calls support evidence gathering, write separation, and pre-write reconsideration. Case-level evidence shows that the escalated workflow preserves fallback plans, binds retrieved records to the correct action, sequences writes, and decomposes multi-entity requests. Airline results extend the same service-control logic to reservation operations.

This paper studies whether the Metabolic Multi-Agent Optimizer (MMAO) can act as a credible outer-loop optimizer for classification model selection. We propose MMAO-Cls, a mixed-space realization in which each agent jointly encodes a binary feature mask and classifier hyperparameters, while private energy, communal budget, role drift, and lifecycle turnover are mapped to the accuracy-complexity tradeoff of wrapper learning. The implementation is strengthened by deriving feature-budget adaptation from feature-information priors and by regularizing validation reward with both subset compactness and train-validation overfitting gap. We evaluate MMAO-Cls on seven standard tabular benchmarks with three seeds each and compare it against RandomSearch, GA-lite, PSO-lite, and an endogenous no-sharing ablation. On the aggregate validation objective, MMAO-Cls ranks second ($0.9433$) behind GA-lite ($0.9446$). On held-out test performance, it reaches mean score $0.8882$, improving over RandomSearch ($0.8808$) and GA-lite ($0.8857$), remaining close to PSO-lite ($0.8874$) and the no-sharing ablation ($0.8900$), while using the most compact mean held-out feature subset among all compared methods (feature ratio $0.4881$). Pairwise tests show that these margins are not yet statistically significant. The resulting claim is therefore conservative: MMAO-Cls supports classification applicability and compact mixed-space search more clearly than it isolates communal sharing as a decisive standalone advantage.

Learning how an environment behaves from interaction is central to building agents that adapt to unfamiliar tasks. World models learned with deep networks are flexible but data-hungry and transfer poorly beyond their training distribution. Program-synthesized world models, written as source code by LLMs and refined through counterexample-guided inductive synthesis (CEGIS), are instead data-efficient and reusable, yet they have been demonstrated mainly on structured-state worlds with a given object vocabulary, and a single program search does not scale to pixel-rendered environments whose object structure must be hypothesized flexibly. We introduce OPINE-World, an LLM agent that learns an object-centric programmatic world model online from interaction. OPINE-World couples two cooperating agents in a loop of hypothesis and test, one acting in the environment and one synthesizing the model in code with replay verification and model-based planning, and it steers exploration with a Bayesian measure of object-type adequacy we call ontology error. We evaluate OPINE-World on ARC-AGI-3, a benchmark for skill-acquisition efficiency in which the object vocabulary, the goal, and the action semantics are withheld. OPINE-World solves 20 of 25 games without per-game training and reaches an action-efficiency score of 78.4 against the human baseline.

Large language models are trained to predict the next token, not to act inside a specific API. In niche enterprise SaaS workflows — where success means hitting the right endpoint with the right nested arguments in the right order — this objective mismatch shows up as silent failures: dropped required fields, hallucinated tools, or early stops after a single read. We ask whether Reinforcement Learning with Verifiable Rewards (RLVR), applied directly in the target environment, closes the gap. As a proof of concept we build a suite of five synthetic environments emulating the Jira REST v3 and Confluence v2 APIs at schema fidelity; rewards are computed entirely from the tool-call trace, with no live API, no learned judge, and no human label in the loop. Scoring prompted Qwen3-1.7B and Qwen3.5-4B on the same checkers that drive GRPO training, we find that on the four scenarios whose rewards are non-degenerate the RL-trained policy lifts average reward from a 4B-baseline range of 0.35--0.92 to 0.95--1.00, with the largest single gain on Confluence page creation ($0.35 \rightarrow 1.00$). We position this as a preliminary step toward outcome-optimised small models for niche enterprise APIs, and foreground two limitations a workshop reader should weigh: hand-crafting verifiable rewards does not scale beyond the handful of endpoints reported here, and one of our five scenarios (ticket-transition) has a saturating reward shape that the prompted 4B already maxes out.

Deep learning (DL) frameworks are critical AI infrastructures that often hide bugs with serious security implications. While dynamic approaches such as fuzzing are effective in uncovering these bugs, they require real test execution and incur high computational costs. Static analysis is a natural complement because it can detect bugs without runtime execution, offering fast and scalable testing. Unfortunately, there is still limited work targeting static analysis for DL frameworks due to their multilingual architectures and tensor-related program state. We present Phoenix, the first LLM-based static analysis technique for DL frameworks. Our key insight is that cross-language tensor flows in DL frameworks can be modeled, together with concrete code context, as a structured semantic bridge intermediate representation (SBIR) that LLMs can analyze for potential bugs in tensor semantic propagation. We implement this insight through a multi-agent workflow. A summarization agent first distills bug summaries from historical bug-fix patches and CWE rules. Guided by each summary, an extraction agent identifies bug-relevant repository symbols for code retrieval, and a generation agent synthesizes grounded SBIRs from the retrieved context. Finally, an analysis agent is leveraged to check SBIRs and report potential bugs. Our evaluation shows that Phoenix is a practical complement to dynamic DL framework testing for bug finding. To date, Phoenix has found 31 real new bugs in PyTorch for different heterogeneous hardware backends (Intel CPU, NVIDIA CUDA, and Apple MPS). Among them, 20 submitted bug-fixing patches have been merged into upstream.

Scientific literature search often requires more than retrieving papers from a single query: users' intents are underspecified, preference-dependent, and evolve through interaction. Existing search agents typically rely on fixed pipelines or implicit language-only reasoning, making their search strategies difficult to control, inspect, and refine. We introduce PaperPilot, a multi-turn literature search agent that frames scientific search as workflow induction. Given an anchor paper and a user query, PaperPilot constructs an executable DAG of paper-search operators, including keyword search, citation expansion, filtering, scoring, reranking, and evidence extraction. User feedback is then used to refine both the query and the workflow itself. We train PaperPilot with supervised workflow imitation and preference optimization over controlled workflow corruptions. Experiments show that PaperPilot-9B improves over the base Qwen3.5-9B toolset agent under multi-turn interaction, increasing Hit@5 from 58.0 to 77.0, MRR from 47.5 to 59.4, and nDCG@10 from 26.8 to 32.5, while reducing workflow execution errors from 9.5% to 0%. These results show that explicit, editable search workflows provide an effective and controllable interface for aligning literature search agents with complex scientific intent.

Hallucination detection for retrieval-augmented generation (RAG) is usually evaluated on natural-language document evidence. However, grounded generation systems increasingly rely on structured inputs: source code, developer-tool output, markdown documents, tables, and repository metadata. We introduce a unified benchmark for span-level hallucination detection over code, tool output, structured documents, and existing natural-language RAG datasets. The benchmark is built by starting from grounded correct answers, injecting localized hallucinations with exact character labels, and validating the code test split with evidence-based review. Our fine-tuned Qwen3.5-2B detector reaches 0.689 span-F1 on the unified test set and 0.60 on the code-agent source, where it substantially outperforms LettuceDetect-large (0.17) and the strongest zero-shot LLM judges we evaluated (at most 0.22). The same model remains competitive on established natural-language benchmarks, with 81.8 RAGTruth example-F1 and 0.724 English PsiloQA IoU.

AI coding agents increasingly rely on skills: structured context bundles, typically a SKILL$.$md file with a YAML header and Markdown body, loaded on demand for domain knowledge, workflows, and scripts. Public registries such as skills$.$sh now host tens of thousands of skills, making them an emerging unit of reuse in agent-based software engineering. Yet skills have largely been viewed as agent capabilities rather than software artefacts whose content and evolution shape agent behaviour. We present the first empirical study of AI agent skills as engineered artefacts that are authored, reused, customised and maintained, across public registries and personal-use repositories. We mined 18,463 skills from skills$.$sh and 23,199 personal-use skills from 5,876 GitHub repositories, identifying 3,709 reuse links. LLM-based classification into SWEBOK knowledge areas (KAs) shows Software Construction dominates alongside a long tail of specialised areas. A thematic analysis of 180 skills identifies six content categories. Qualitative coding of 444 modifications reveals six themes, of which reworking operational specifications and adapting knowledge and resources are the primary target of change. Our findings show that reuse is largely a one-time copy operation: most reused skills remain near-verbatim, 53% are never modified after adoption, and subsequent local maintenance is overwhelmingly additive. Customisation primarily adapts skills to local environments, whereas evolution accretes new inline domain knowledge. Across both, a stable behavioural contract - how a skill interacts with users, monitors runtime state, and recovers from failures - remains almost untouched. These results suggest maintenance effort should focus on project-specific bindings, and that registries and tool support should enable consolidating the domain knowledge skills re-author in isolation.