arXiv subjects

Zhonghao Jiang

Publications and source records attributed to Zhonghao Jiang.

9 recordsLinked to original sources

arXiv2026

How Agent Skills Evolve: A Large-Scale Empirical Study of Skills on GitHub

Large language model-based agents increasingly rely on external skills to reuse task-specific knowledge without retraining the model. As skills become common for extending agent capabilities, understanding creation and evolution is critical for reliable agent systems. However, existing studies mainly examine agent skills from a static perspective, leaving their evolution in real-world repositories largely unexplored. This limits our understanding of the patterns and motivations behind skill changes. To fill this gap, we present a large-scale empirical study of agent skills, capturing 84,200 skills and 750,898 commits from GitHub. We manually construct taxonomies of skill components, change patterns, and evolution motivations, and build an automated annotation pipeline to label changes at scale. Based on the taxonomy, we conduct quantitative analyses of skill composition and post-creation evolution frequency, together with qualitative assessments of modified components, change patterns, and motivations. Our findings reveal five implications: (i) reliability receives less attention in skill composition and change, motivating execution-contract skill design that incorporates verification and recovery mechanisms; (ii) evolution changes concentrate in project-coupled skills, calling for automated just-in-time skill updating mechanisms; (iii) change patterns are dominated by incremental enrichment and refinement, motivating pattern-guided skill enhancers that diagnose and improve incomplete skill components; (iv) redundancy is a practical maintenance issue, highlighting the need for context-efficient skill representation; (v) security-motivated changes show limited risk reduction, suggesting skill regression testing methods that objectively verify skill functionality and security. Beyond these insights, our dataset provides a basis for future studies on automated skill techniques.

cs.SE↗

arXiv2026

Well-Specified and Well-Tested? Assessing SWE-bench-Style Datasets with a Self-Evolving Agent

Large language models (LLMs) have shown strong performance on software engineering (SE) tasks, driving the need for high-quality datasets for reliable training and evaluation. SWE-bench-style datasets (SWE datasets) are widely used to evaluate LLM-based coding agents on real-world software issues, where task clarity and test coverage directly affect whether instances are solvable and verifiable. However, quality issues persist in existing SWE datasets, and manual annotation is costly and labor-intensive, motivating LLM-based annotation. Existing approaches typically rely on static, overly general rubrics, limiting their ability to handle diverse and nuanced scenarios and often yielding high false positive rates. In contrast, human annotators draw on accumulated experience to make context-aware judgments. Inspired by this, we propose ExLabel, an LLM-based, self-evolving, experience-guided assistant for task clarity and test coverage assessment in SWE datasets. ExLabel maintains an evolving experience bank that distills cross-instance knowledge into stable yet adaptive judgment criteria and reuses relevant experience to guide the assessment of new instances. For systematic evaluation, we construct SWE-TQ, a new benchmark spanning eight programming languages, by cleaning human annotations from SWE-bench and Multi-SWE-bench. Experiments with three backbone LLMs show that ExLabel achieves the best F1 across all settings, with relative improvements of 44.2%/49.9% and 24.0%/18.4% over the best baseline for task clarity and test coverage assessment on the Python/Multilingual subsets, respectively. Moreover, a coding model fine-tuned on SWE-rebench data filtered by ExLabel resolves 66.2% of issues on SWE-bench Verified, outperforming the model trained on data filtered by the best baseline by 14.5%.

cs.SE↗

arXiv2026

From Faulty Claims to Incorrect Patches: An Empirical Study of Task-Grounded Reasoning in Coding Agents

Large language model (LLM)-based coding agents increasingly resolve real-world software engineering tasks, yet still fail on many of them. Prior trajectory studies characterize these failures through task-agnostic patterns of agent behavior, such as repeated actions. However, an agent may follow a sound workflow yet act on a faulty understanding of the existing code, the task requirements, or a candidate fix. Reasoning models, which power most state-of-the-art coding agents, reveal this understanding in their native chain-of-thought (CoT), whose claims about the task can be checked against the facts. Yet it remains unclear how many such checkable claims the CoT contains, how many are faulty, and whether faulty claims actually cause incorrect patches. We present the first empirical study of such claims, which we term task-grounded reasoning, analyzing 400 trajectories of four open-weight reasoning models on SWE-bench Pro. We extract atomic claims from the CoT with an LLM, assess them against evidence gathered by an LLM agent, and manually trace the defects of incorrect patches back to faulty claims, confirming each link with a claim-corrected replay. Task-grounded reasoning covers only 18.5% of the CoT tokens, yet yields 191 atomic claims per trajectory, at least 9.6% of which are faulty. Faulty claims also occur in 96.7% of the resolved trajectories, so their mere presence is a weak signal of failure. Yet two thirds of the 188 incorrect patches trace back to faulty claims. Correcting such a claim makes the agent fix the linked defect in 89.0% of the cases, and correcting all such claims resolves 11.7% of the failed tasks. Our findings show that the CoT carries task-specific signals for diagnosing and repairing agent failures, but exploiting them requires pinpointing the few faulty claims that actually cause patch defects.

cs.SE↗

arXiv2026

Agentic Software Issue Resolution with Large Language Models: A Survey

Software issue resolution aims to address real-world issues in software repositories based on natural language descriptions provided by users, and represents a key aspect of software maintenance. With the rapid development of large language models (LLMs) in reasoning and generation, LLM-based approaches have made significant progress in automated software issue resolution. However, resolving real-world software issues is inherently complex and requires long-horizon reasoning, iterative exploration, and feedback-driven decision-making, which demand agentic capabilities beyond conventional single-step approaches. Recently, LLM-based agentic systems have emerged as a promising research direction for software issue resolution, accompanied by rapid growth in the relevant literature. Advances in agentic software issue resolution can not only greatly improve the efficiency and quality of software maintenance, but also provide a realistic environment for evaluating the reasoning, planning, and execution capabilities of agentic systems, thereby bridging artificial intelligence and software engineering. This work presents a systematic survey of 242 recent studies at the forefront of research on LLM-based agentic software issue resolution. It outlines the general workflow of the task and establishes a taxonomy across three dimensions: benchmarks, techniques, and empirical studies. Furthermore, it highlights reinforcement learning as an increasingly important training paradigm for agentic systems in software engineering. Finally, it summarizes key challenges and outlines promising directions for future research.

cs.SE↗

arXiv2026

Doc2Feat-Bench: Evaluating Documentation-Driven Feature Addition

Documentation changes in mature software projects often describe newly introduced or modified behavior. This makes them a natural basis for documentation-driven feature addition, where software engineering agents implement features from public-facing documentation updates. However, existing benchmarks for agentic software development primarily rely on issue reports written for maintainer coordination and often contain implementation-level details. Motivated by this observation, this work introduces Doc2Feat-bench, a benchmark for documentation-driven feature addition. Doc2Feat-bench consists of 634 tasks across 10 mature open-source software projects, involving about 114k code changes in total. Each task pairs a documentation change with the corresponding implementation and developer-written tests, enabling evaluation of whether a software engineering agent can implement the feature described by the documentation diff. The benchmark is constructed through a five-phase pipeline, starting from release notes, and offers broad coverage across libraries, developer tools, and frameworks while preserving real-world development settings. To facilitate lightweight and reliable evaluation under limited resources, we further curate a human-validated subset named Doc2Feat-bench Verified. It contains 114 high-quality instances whose task clarity and evaluation validity are manually verified. We use Doc2Feat-bench to assess a range of state-of-the-art software engineering agents with two widely used scaffolds. Experimental results show that despite significant token consumption, the best task success rate remains as low as 37.72% (OpenHands with Qwen3-Coder-480B). Our analysis reveals that software engineering agents face key challenges in performing cross-file edits, understanding existing code modules, and correctly interpreting documentation semantics.

cs.SE↗

arXiv2026

Effective and Efficient Context Retrieval via Partial Dependency Graph for Repository-Level Code Generation

LLM-based repository-level code generation aims to generate code using the context available in a software repository, requiring LLMs to reason over complex code dependencies. Due to limited context windows and insufficient repository-specific understanding, LLMs typically rely on retrieval-augmented generation (RAG) to incorporate relevant code. Early RAG approaches primarily employ similarity-based retrieval, which often fails to retrieve code snippets that the target function depends on. Recent work introduces graph-based retrieval to model such dependencies, but typically relies on manually designed rules and static global graphs, leading to limited flexibility and high construction and maintenance costs. In contrast, human developers collect helpful context by implicitly constructing a partial dependency graph and iteratively inspecting along it. Inspired by this behavior, we propose DyRetriever, an efficient context retrieval method via partial dependency graphs. DyRetriever uses an LLM to first select a set of entry-point functions and then perform multi-hop reasoning along the code dependency graph. During multi-hop reasoning, it uses the LLM's semantic understanding to validate whether a function can help generate the target function, eliminating manually designed rules and enabling flexibility across scenarios. Instead of statically constructing a global dependency graph, DyRetriever builds a partial graph on demand and discards it after use, reducing construction and maintenance costs. We integrate DyRetriever with a similarity-based code retriever to build DyCoder and evaluate it on CoderEval and DevEval. Experimental results show that DyCoder achieves relative Pass@1 improvements of 25.63% and 59.73% on CoderEval and DevEval, respectively, compared with existing RAG-based methods, while being 7.4x faster than baselines based on static dependency graph construction.

cs.SE↗

arXiv2025

Issue Localization via LLM-Driven Iterative Code Graph Searching

Issue solving aims to generate patches to fix reported issues in real-world code repositories according to issue descriptions. Issue localization forms the basis for accurate issue solving. Recently, LLM-based issue localization methods have demonstrated state-of-the-art performance. However, these methods either search from files mentioned in issue descriptions or in the whole repository and struggle to balance the breadth and depth of the search space to converge on the target efficiently. Moreover, they allow LLM to explore whole repositories freely, making it challenging to control the search direction to prevent the LLM from searching for incorrect targets. This paper introduces CoSIL, an LLM-driven, powerful function-level issue localization method without training or indexing. CoSIL employs a two-phase code graph search strategy. It first conducts broad exploration at the file level using dynamically constructed module call graphs, and then performs in-depth analysis at the function level by expanding the module call graph into a function call graph and executing iterative searches. To precisely control the search direction, CoSIL designs a pruner to filter unrelated directions and irrelevant contexts. To avoid incorrect interaction formats in long contexts, CoSIL introduces a reflection mechanism that uses additional independent queries in short contexts to enhance formatted abilities. Experiment results demonstrate that CoSIL achieves a Top-1 localization accuracy of 43.3\% and 44.6\% on SWE-bench Lite and SWE-bench Verified, respectively, with Qwen2.5-Coder-32B, average outperforming the state-of-the-art methods by 96.04\%. When CoSIL is integrated into an issue-solving method, Agentless, the issue resolution rate improves by 2.98\%--30.5\%.

cs.SE↗

arXiv2025

AgentDroid: A Multi-Agent Framework for Detecting Fraudulent Android Applications

With the increasing prevalence of fraudulent Android applications such as fake and malicious applications, it is crucial to detect them with high accuracy and adaptability. We present AgentDroid, a novel tool for Android fraudulent application detection based on multi-modal analysis and multi-agent systems. AgentDroid overcomes the limitations of traditional detection methods such as the inability to handle multimodal data and high false alarm rates. It processes Android applications and extracts a series of multi-modal data for analysis. Multiple LLM-based agents with specialized roles analyze the relevant data and collaborate to detect complex fraud effectively. We curated a dataset containing various categories of fraudulent applications and legitimate applications and validated our tool on this dataset. Experimental results indicate that our multi-agent tool based on GPT-4o achieves an accuracy of 91.7% and an F1-Score of 91.68%, outperforming the baseline methods. A video of AgentDroid is available at https://youtu.be/YOM9Ex-nBts.

cs.SE↗

arXiv2024

DFEPT: Data Flow Embedding for Enhancing Pre-Trained Model Based Vulnerability Detection

Software vulnerabilities represent one of the most pressing threats to computing systems. Identifying vulnerabilities in source code is crucial for protecting user privacy and reducing economic losses. Traditional static analysis tools rely on experts with knowledge in security to manually build rules for operation, a process that requires substantial time and manpower costs and also faces challenges in adapting to new vulnerabilities. The emergence of pre-trained code language models has provided a new solution for automated vulnerability detection. However, code pre-training models are typically based on token-level large-scale pre-training, which hampers their ability to effectively capture the structural and dependency relationships among code segments. In the context of software vulnerabilities, certain types of vulnerabilities are related to the dependency relationships within the code. Consequently, identifying and analyzing these vulnerability samples presents a significant challenge for pre-trained models. In this paper, we propose a data flow embedding technique to enhance the performance of pre-trained models in vulnerability detection tasks, named DFEPT, which provides effective vulnerability data flow information to pre-trained models. Specifically, we parse data flow graphs from function-level source code, and use the data type of the variable as the node characteristics of the DFG. By applying graph learning techniques, we embed the data flow graph and incorporate relative positional information into the graph embedding using sine positional encoding to ensure the completeness of vulnerability data flow information. Our research shows that DFEPT can provide effective vulnerability semantic information to pre-trained models, achieving an accuracy of 64.97% on the Devign dataset and an F1-Score of 47.9% on the Reveal dataset.

cs.SE↗