arXiv subjects

Bihuan Chen

Publications and source records attributed to Bihuan Chen.

At least 19 recordsLinked to original sources

arXiv2026

SkillSentry: Reliable Skill Execution for LLM Agents via Runtime Assurance

LLM agents are increasingly equipped with skills to perform complex tasks through multi-step reasoning and tool use. Although skills provide reusable procedural knowledge, agents may still execute them unreliably. Even when an agent has demonstrated the capability to complete tasks under the guidance of a skill, it may fail to do so consistently across similar tasks or repeated runs due to deviations from the skill procedure or incorrect execution of individual steps. Such instability limits the practical reliability of LLM agents. To address this problem, we propose SkillSentry, a skill-oriented runtime assurance framework built upon a new domain-specific language (DSL) for representing runtime guidance for skill execution. SkillSentry initializes the runtime guidance by combining a skill specification extracted from the corresponding skill document with execution experience mined from historical successful and failed traces. It then wraps around the agent execution loop to monitor and guide skill execution under the current guidance, while iteratively refining the guidance using newly collected traces. We evaluate SkillSentry on 15 skills across two LLM agents, each paired with two backbone models, i.e., Claude Code with Claude-Haiku-4.5 and Claude-Opus-4.6, and Codex with GPT-5.2 and GPT-5.4. Our results show that SkillSentry improves the task success rate of LLM agents by 24.1% across skills, on average, while exhibiting lower variability across repeated runs.

cs.AI↗

arXiv2026

DOCSCHISEL: Adaptive Tool Documentation Optimization Framework for LLM Agents

Large language models (LLMs) increasingly rely on external tools to accomplish complex real-world tasks, making tool documentation a critical grounding resource for LLM agents. Existing studies mainly focus on improving the tool-use capabilities of LLM agents, while largely treating tool documentation as a fixed input. Although several recent works attempt to optimize tool documentation through rewriting or compression, little is known about how the information contained in tool documentation affects agent performance across different settings. To bridge this gap, we conduct a large-scale empirical study on tool documentation for LLM agents. Our study reveals substantial heterogeneity in the information fields provided by existing tool documentation. Moreover, the effectiveness of different information fields is highly dependent on the task domain, LLM backbone, and agent paradigm, indicating that no fixed tool documentation can consistently generalize across diverse agent settings. Motivated by these findings, we propose DocsChisel, an adaptive tool documentation optimization framework for LLM agents. DocsChisel analyzes failed execution traces of a target LLM agent to identify documentation-related issues, and iteratively optimizes tool documentation by adding, removing, and refining information fields for each tool. We evaluate DocsChisel against two state-of-the-art baselines, i.e., EasyTool and DRAFT. Experimental results show that DocsChisel improves the task success rate of LLM agents by 95.89% over the original tool documentation and by 75.15%, on average, over existing baselines, while incurring limited optimization time and token overhead

cs.LG↗

arXiv2026

FlowScout: From Execution Feedback to Reliable Tool-Using Agent Workflows

Agentic workflows have become an important abstraction for building reliable LLM-based automation systems by organizing large language models (LLMs), tools, and control logic into explicit execution structures. However, constructing high-quality agentic workflows remains largely manual and requires substantial domain expertise. Recent studies have explored automatic agentic workflow generation from historical task-solving records, but they mainly produce LLM-centric workflows, where real tool executions are abstracted and simulated by LLM nodes, limiting the usability and stability of generated workflows. To address these limitations, we propose FlowScout, an execution-guided framework for generating tool-integrated agentic workflows from historical task-solving records. Specifically, FlowScout represents an agentic workflow as a directed graph composed of LLM nodes, tool-calling nodes, and dependency edges. It first mines a common tool coordination skeleton from historical records to construct an initial workflow, and then refines the workflow topology through Monte Carlo tree search guided by execution feedback. We evaluate FlowScout on four representative task domains and compare it with three baselines, i.e., PM4Py, ReAct and AFlow. Experimental results show that agentic workflows generated by FlowScout improve tool invocation correctness by at least 92.69% and execution quality by at least 17.66% over the baselines, while achieving lower performance variation across repeated runs.

cs.LG↗

arXiv2026

ProfMalPlus: Agent-Coordinated Detection of Malicious NPM Packages via Static-Dynamic Analysis Synergy

Open source software is vulnerable to supply-chain attacks through transitive dependencies, especially malicious code injected into NPM packages. Existing detectors often inadequately model obfuscated behavior, overlook JavaScript's object-centric features, poorly coordinate static and dynamic analysis, and lose semantic information during behavior abstraction. We propose ProfMalPlus, a malicious NPM package detector combining object-sensitive behavior graphs with coordinated LLM reasoning over annotated code slices. It identifies installation commands and entry files, then constructs graphs capturing sensitive APIs, third-party calls, and unresolved calls. From these graphs, ProfMalPlus extracts security-relevant slices and adds inline static analysis evidence. Local judge agents independently assess each slice. Self-consistency consolidates repeated judgements to reduce LLM variance, while a global judge synthesizes their reports into an entry-level verdict. For undetermined cases, a router selects either third-party enrichment, which adds registry derived module and method semantics, or dynamic augmentation, which executes the package in a sandbox to resolve runtime dependent behavior. The enriched evidence is fed back for reassessment. Finally, a localization agent reports malicious code snippets with explanations. ProfMalPlus achieves a 98.1% F1-score, outperforming state-of-the-art detectors by 3.5% to 52.6%. It also identified 597 previously unknown malicious packages, all confirmed and removed from NPM.

cs.SE↗

arXiv2026

Arachne: Orchestrating Cascades for Efficient Text-to-Video Model Training

The rising demand for AI-generated videos is fueled by advances in large-scale Text-to-Video (T2V) models, trained on extensive datasets of video clips spanning diverse resolutions and durations. To address this data heterogeneity, current training methods often use a bucketing strategy that groups samples into discrete buckets for efficiency. However, this approach struggles to scale with compute and data volumes under static parallelism schemes, such as data and sequence parallelism, leading to significant workload imbalances and hardware under-utilization. In this paper, we present Arachne, a novel training framework for efficient T2V model training at scale. Arachne decomposes the training process into fine-grained computational units, called \textit{cascades}, orchestrating their distributed execution and synchronization across the cluster through coordinated spatial and temporal optimization. Our comprehensive evaluation demonstrates that Arachne reduces iteration time by up to 65\% over leading frameworks, exhibiting a positive scaling trend where its performance advantages amplify as training scale grows.

cs.DC↗

arXiv2026

Bridging Generation and Training: A Systematic Review of Quality Issues in LLMs for Code

Large language models (LLMs) frequently generate defective outputs in code generation tasks, ranging from logical bugs to security vulnerabilities. While these generation failures are often treated as model-level limitations, empirical evidence increasingly traces their root causes to imperfections within the training corpora. Yet, the specific mechanisms linking training data quality issues to generated code quality issues remain largely unmapped. This paper presents a systematic literature review of 114 primary studies to investigate how training data quality issues propagate into code generation. We establish a unified taxonomy that categorizes generated code quality issues across nine dimensions and training data quality issues into code and non-code attributes. Based on this taxonomy, we formalize a causal framework detailing 18 typical propagation mapping mechanisms. Furthermore, we synthesize state-of-the-art detection and mitigation techniques across the data, model, and generation lifecycles. The reviewed literature reveals a clear methodological shift: quality assurance is transitioning from reactive, heuristic-based post-generation filtering toward proactive, data-centric governance and closed-loop repair. Finally, we identify open challenges and outline research directions for developing reliable LLMs for code through integrated data curation and continuous evaluation. Our repository is available at https://github.com/SYSUSELab/From-Data-to-Code.

cs.SE↗

arXiv2026

VulWeaver: Weaving Broken Semantics for Grounded Vulnerability Detection

Detecting vulnerabilities in source code remains critical yet challenging, as conventional static analysis tools construct inaccurate program representations, while existing LLM-based approaches often miss essential vulnerability context and lack grounded reasoning. In this paper, we introduce VulWeaver, a novel LLM-based approach that weaves broken program semantics into accurate representations and extracts holistic vulnerability context for grounded vulnerability detection. VulWeaver first constructs an enhanced unified dependency graph (UDG) by integrating deterministic rules with LLM-based semantic inference to address static analysis inaccuracies. It then extracts holistic vulnerability context by combining explicit contexts from program slicing with implicit contexts, including usage, definition, and declaration information. Finally, VulWeaver employs meta-prompting with vulnerability type specific expert guidelines to steer LLMs through systematic reasoning, aggregated via majority voting for robustness. Extensive experiments on PrimeVul4J dataset show that VulWeaver achieves a precision of 0.82, recall of 0.71, and F1-score of 0.76, outperforming state-of-the-art learning-based, LLM-based, and agent-based baselines by 25%, 17%, and 21% in F1-score, respectively. Notably, VulWeaver attains a VP-S score of 0.58, 164% higher than the best baseline, confirming its strong discriminative power

cs.SE↗

arXiv2026

From Component Manipulation to System Compromise: Understanding and Detecting Malicious MCP Servers

The model context protocol (MCP) standardizes how LLMs connect to external tools and data sources, enabling faster integration but introducing new attack vectors. Despite the growing adoption of MCP, existing MCP security studies classify attacks by their observable effects, obscuring how attacks behave across different MCP server components and overlooking multi-component attack chains. Meanwhile, existing defenses are less effective when facing multi-component attacks or previously unknown malicious behaviors. This work presents a component-centric perspective for understanding and detecting malicious MCP servers. First, we build the first component-centric PoC dataset of 114 malicious MCP servers where attacks are achieved as manipulation over MCP components and their compositions. We evaluate these attacks' effectiveness across two MCP hosts and five LLMs, and uncover that (1) component position shapes attack success rate; and (2) multi-component compositions often outperform single-component attacks by distributing malicious logic. Second, we propose and implement Connor, a two-stage behavioral deviation detector for malicious MCP servers. It first performs pre-execution analysis to detect malicious shell commands and extract each tool's function intent, and then conducts step-wise in-execution analysis to trace each tool's behavioral trajectories and detect deviations from its function intent. Evaluation on our curated dataset indicates that Connor achieves an F1-score of 94.6%, outperforming the state of the art by 8.9% to 59.6%. In real-world detection, Connor identifies two malicious servers.

cs.CR↗

arXiv2026

AutoMerge: Search-Based Model Merging Framework for Effective Model Reuse

Software reuse has long been recognized as a critical and widely studied topic in software engineering, offering substantial benefits in reducing development costs, improving software quality, and enhancing operational efficiency. This paradigm extends into deep learning through model reuse. Recently, model merging has emerged in the domain of large language models (LLMs) as a training-free approach that takes multiple task-specific models with the same architecture as source models and merges them without retraining, enhancing model reuse within LLMs. However, no prior work has systematically investigated whether such an approach can be effectively applied to other deep learning models with different architectures across domains. To bridge this gap, we present the first systematic study that evaluates five model merging techniques on three distinct model architectures across three domains: LLMs, image classification, and autonomous driving. Our findings reveal that directly applying existing model merging techniques leads to highly inconsistent results and falls notably short of their success within LLMs. Moreover, a single model merging technique often fails to handle the heterogeneous structural properties within a model, limiting its applicability to different model architectures across domains. Furthermore, the effectiveness of model merging techniques is highly sensitive to hyperparameter configurations, thereby constraining their potential for broader adoption. Inspired by these insights, we propose AutoMerge, a novel search-based model merging framework that first segments complex models into multiple heterogeneous blocks and then systematically explores the merging space to identify the merging technique and its hyperparameter configuration.

cs.SE↗

arXiv2025

ExpertAD: Enhancing Autonomous Driving Systems with Mixture of Experts

Recent advancements in end-to-end autonomous driving systems (ADSs) underscore their potential for perception and planning capabilities. However, challenges remain. Complex driving scenarios contain rich semantic information, yet ambiguous or noisy semantics can compromise decision reliability, while interference between multiple driving tasks may hinder optimal planning. Furthermore, prolonged inference latency slows decision-making, increasing the risk of unsafe driving behaviors. To address these challenges, we propose ExpertAD, a novel framework that enhances the performance of ADS with Mixture of Experts (MoE) architecture. We introduce a Perception Adapter (PA) to amplify task-critical features, ensuring contextually relevant scene understanding, and a Mixture of Sparse Experts (MoSE) to minimize task interference during prediction, allowing for effective and efficient planning. Our experiments show that ExpertAD reduces average collision rates by up to 20% and inference latency by 25% compared to prior methods. We further evaluate its multi-skill planning capabilities in rare scenarios (e.g., accidents, yielding to emergency vehicles) and demonstrate strong generalization to unseen urban environments. Additionally, we present a case study that illustrates its decision-making process in complex driving scenarios.

cs.RO↗

arXiv2025

Argus: Resilience-Oriented Safety Assurance Framework for End-to-End ADSs

End-to-end autonomous driving systems (ADSs), with their strong capabilities in environmental perception and generalizable driving decisions, are attracting growing attention from both academia and industry. However, once deployed on public roads, ADSs are inevitably exposed to diverse driving hazards that may compromise safety and degrade system performance. This raises a strong demand for resilience of ADSs, particularly the capability to continuously monitor driving hazards and adaptively respond to potential safety violations, which is crucial for maintaining robust driving behaviors in complex driving scenarios. To bridge this gap, we propose a runtime resilience-oriented framework, Argus, to mitigate the driving hazards, thus preventing potential safety violations and improving the driving performance of an ADS. Argus continuously monitors the trajectories generated by the ADS for potential hazards and, whenever the EGO vehicle is deemed unsafe, seamlessly takes control through a hazard mitigator. We integrate Argus with three state-of-the-art end-to-end ADSs, i.e., TCP, UniAD and VAD. Our evaluation has demonstrated that Argus effectively and efficiently enhances the resilience of ADSs, improving the driving score of the ADS by up to 150.30% on average, and preventing up to 64.38% of the violations, with little additional time overhead.

cs.AI↗

arXiv2025

TigAug: Data Augmentation for Testing Traffic Light Detection in Autonomous Driving Systems

Autonomous vehicle technology has been developed in the last decades with recent advances in sensing and computing technology. There is an urgent need to ensure the reliability and robustness of autonomous driving systems (ADSs). Despite the recent achievements in testing various ADS modules, little attention has been paid on the automated testing of traffic light detection models in ADSs. A common practice is to manually collect and label traffic light data. However, it is labor-intensive, and even impossible to collect diverse data under different driving environments. To address these problems, we propose and implement TigAug to automatically augment labeled traffic light images for testing traffic light detection models in ADSs. We construct two families of metamorphic relations and three families of transformations based on a systematic understanding of weather environments, camera properties, and traffic light properties. We use augmented images to detect erroneous behaviors of traffic light detection models by transformation-specific metamorphic relations, and to improve the performance of traffic light detection models by retraining. Large-scale experiments with four state-of-the-art traffic light detection models and two traffic light datasets have demonstrated that i) TigAug is effective in testing traffic light detection models, ii) TigAug is efficient in synthesizing traffic light images, and iii) TigAug generates traffic light images with acceptable naturalness.

cs.SE↗

arXiv2025

Lifting the Veil on Composition, Risks, and Mitigations of the Large Language Model Supply Chain

Large language models (LLMs) have sparked significant impact with regard to both intelligence and productivity. Numerous enterprises have integrated LLMs into their applications to solve their own domain-specific tasks. However, integrating LLMs into specific scenarios is a systematic process that involves substantial components, which are collectively referred to as the LLM supply chain. A comprehensive understanding of LLM supply chain composition, as well as the relationships among its components, is crucial for enabling effective mitigation measures for different related risks. While existing literature has explored various risks associated with LLMs, there remains a notable gap in systematically characterizing the LLM supply chain from the dual perspectives of contributors and consumers. In this work, we develop a structured taxonomy encompassing risk types, risky actions, and corresponding mitigations across different stakeholders and components of the supply chain. We believe that a thorough review of the LLM supply chain composition, along with its inherent risks and mitigation measures, would be valuable for industry practitioners to avoid potential damages and losses, and enlightening for academic researchers to rethink existing approaches and explore new avenues of research.

cs.SE↗

arXiv2025

Vul-RAG: Enhancing LLM-based Vulnerability Detection via Knowledge-level RAG

Although LLMs have shown promising potential in vulnerability detection, this study reveals their limitations in distinguishing between vulnerable and similar-but-benign patched code (only 0.06 - 0.14 accuracy). It shows that LLMs struggle to capture the root causes of vulnerabilities during vulnerability detection. To address this challenge, we propose enhancing LLMs with multi-dimensional vulnerability knowledge distilled from historical vulnerabilities and fixes. We design a novel knowledge-level Retrieval-Augmented Generation framework Vul-RAG, which improves LLMs with an accuracy increase of 16% - 24% in identifying vulnerable and patched code. Additionally, vulnerability knowledge generated by Vul-RAG can further (1) serve as high-quality explanations to improve manual detection accuracy (from 60% to 77%), and (2) detect 10 previously-unknown bugs in the recent Linux kernel release with 6 assigned CVEs.

cs.SE↗

arXiv2025

A Survey of Fuzzing Open-Source Operating Systems

Vulnerabilities in open-source operating systems (OSs) pose substantial security risks to software systems, making their detection crucial. While fuzzing has been an effective vulnerability detection technique in various domains, OS fuzzing (OSF) faces unique challenges due to OS complexity and multi-layered interaction, and has not been comprehensively reviewed. Therefore, this work systematically surveys the state-of-the-art OSF techniques, categorizes them based on the general fuzzing process, and investigates challenges specific to kernel, file system, driver, and hypervisor fuzzing. Finally, future research directions for OSF are discussed.

cs.OS↗

arXiv2024

DynNPC: Finding More Violations Induced by ADS in Simulation Testing through Dynamic NPC Behavior Generation

Recently, a number of simulation testing approaches have been proposed to generate diverse driving scenarios for autonomous driving systems (ADSs) testing. However, the behaviors of NPC vehicles in these scenarios generated by previous approaches are predefined and mutated before simulation execution, ignoring traffic signals and the behaviors of the Ego vehicle. Thus, a large number of the violations they found are induced by unrealistic behaviors of NPC vehicles, revealing no bugs of ADSs. Besides, the vast scenario search space of NPC behaviors during the iterative mutations limits the efficiency of previous approaches. To address these limitations, we propose a novel scenario-based testing framework, DynNPC, to generate more violation scenarios induced by the ADS. Specifically, DynNPC allows NPC vehicles to dynamically generate behaviors using different driving strategies during simulation execution based on traffic signals and the real-time behavior of the Ego vehicle. We compare DynNPC with state-of-the-art scenario-based testing approaches. Our evaluation has demonstrated the effectiveness and efficiency of DynNPC in finding more violation scenarios induced by the ADS.

cs.SE↗

arXiv2024

RoadGen: Generating Road Scenarios for Autonomous Vehicle Testing

With the rapid development of autonomous vehicles, there is an increasing demand for scenario-based testing to simulate diverse driving scenarios. However, as the base of any driving scenarios, road scenarios (e.g., road topology and geometry) have received little attention by the literature. Despite several advances, they either generate basic road components without a complete road network, or generate a complete road network but with simple road components. The resulting road scenarios lack diversity in both topology and geometry. To address this problem, we propose RoadGen to systematically generate diverse road scenarios. The key idea is to connect eight types of parameterized road components to form road scenarios with high diversity in topology and geometry. Our evaluation has demonstrated the effectiveness and usefulness of RoadGen in generating diverse road scenarios for simulation.

cs.SE↗

arXiv2024

Scalable and Precise Application-Centered Call Graph Construction for Python

Call graph construction is the foundation of inter-procedural static analysis. PYCG is the state-of-the-art approach for constructing call graphs for Python programs. Unfortunately, PyCG does not scale to large programs when adapted to whole-program analysis where application and dependent libraries are both analyzed. Moreover, PyCG is flow-insensitive and does not fully support Python's features, hindering its accuracy. To overcome these drawbacks, we propose a scalable and precise approach for constructing application-centered call graphs for Python programs, and implement it as a prototype tool JARVIS. JARVIS maintains a type graph (i.e., type relations of program identifiers) for each function in a program to allow type inference. Taking one function as an input, JARVIS generates the call graph on-the-fly, where flow-sensitive intra-procedural analysis and inter-procedural analysis are conducted in turn and strong updates are conducted. Our evaluation on a micro-benchmark of 135 small Python programs and a macro-benchmark of 6 real-world Python applications has demonstrated that JARVIS can significantly improve PYCG by at least 67% faster in time, 84% higher in precision, and at least 20% higher in recall.

cs.SE↗