As APIs across standard libraries and software ecosystems evolve, the API knowledge frozen in pre-trained LLMs becomes outdated. Prompt-based methods supply updated documentation at inference time yet suffer from context neglect, often reverting to memorized deprecated usages. Existing parametric approaches, whether fine-tuning or model editing, face two common challenges: they treat API changes as textual fact updates, overlooking the changes in code semantics needed for executable correctness (C1), and lack a mechanism to concentrate updates on positions involved in the API evolution, causing parameter interference beyond the evolution boundary (C2). We propose AEG-Edit, which constructs a heterogeneous API evolution graph, encodes and aligns the graph signal with the LLM hidden space, and concentrates gradient effort on positions involved in the API evolution through API-weighted windowing and optimization. On RustEvo²+ (708 Rust tasks) and PyEvo (622 Python tasks) across two representative open-source LLMs at the 7–8B parameter scale, AEG-Edit achieves statistically significant Pass@1 gains on RustEvo²+, improving over the strongest baseline by up to 6.49% on Qwen2.5-7B-Instruct. Ablation and transferability analyses confirm that the heterogeneous API evolution graph drives the gains and transfers to alternative editing backbones.

Fine-tuning safety-aligned large language models (LLMs) can substantially compromise their safety. Previous approaches require many safety samples or calibration sets, which not only incur significant computational overhead during realignment but also lead to noticeable degradation in model utility. Contrary to this belief, we show that safety alignment can be fully recovered with only a single safety example, without sacrificing utility and at minimal cost. Remarkably, this recovery is effective regardless of the number of harmful examples used in fine-tuning or the size of the underlying model, and convergence is achieved within just a few epochs. Furthermore, we uncover the low-rank structure of the safety gradient, which explains why such efficient correction is possible. We validate our findings across five safety-aligned LLMs and multiple datasets, demonstrating the generality of our approach.

The Internet is transitioning from Web3 toward Web4, where autonomous agents serve as independent economic actors. These agents can now hold crypto wallets, execute on-chain trades, and pay for external API calls. This transition calls for a new infrastructure stack capable of supporting key agent operations, including agent-to-tool interaction, agent-to-agent payments, and verifiable agent identity, represented by emerging protocols such as the Model Context Protocol, x402, and EIP-8004. Despite growing industrial interest in these protocols, the real-world Web4 agent ecosystem remains largely underexplored. To bridge this gap, we conduct the first large-scale empirical study of the Web4 ecosystem. Specifically, our study targets three interconnected questions: how Web4 agents are deployed and used in practice; what engineering challenges developers face when building Web4 agents; how current project communities respond to these challenges. To answer these questions, we analyze 99,448 multi-chain identity registrations, 317,596,323 transaction logs, the source code of 341 MCP projects, and 349 filtered GitHub issues. Our findings reveal that autonomous agents have established a highly active machine-to-machine payment economy, processing millions of daily transactions. However, this growth is built on immature infrastructure, including identity/authorization practice, cross-environment operation, and payment interoperability. Our follow-up analysis shows that community responses are visible but unevenly distributed across repositories, and payment interoperability remains the most persistent unresolved bottleneck. Overall, this study reveals a critical gap between the rapid growth of the Web4 agent economy and its fragile underlying infrastructure, highlighting future directions for building a more secure Web4 agent ecosystem.

Large Language Models (LLMs) have demonstrated remarkable capabilities in code generation tasks. However, repository-level code generation presents unique challenges, particularly due to the need to utilize information spread across multiple files within a repository. Specifically, successful generation depends on a solid grasp of both general, context-agnostic knowledge and specific, context-dependent knowledge. While LLMs are widely used for the context-agnostic aspect, existing retrieval-based approaches sometimes fall short as they are limited in obtaining a broader and deeper repository context. In this article, we present CatCoder , a novel code generation framework designed for statically typed programming languages. CatCoder enhances repository-level code generation by integrating relevant c ode a nd t ype context. Specifically, it leverages static analyzers to extract type dependencies and merges this information with retrieved code to create comprehensive prompts for LLMs. To evaluate the effectiveness of CatCoder , we adapt and construct benchmarks that include 199 Java tasks and 90 Rust tasks. The results show that CatCoder outperforms the RepoCoder baseline by up to 14.44% and 17.35%, in terms of compile@ \( k \) and pass@ \( k \) scores. In addition, the generalizability of CatCoder is assessed using various LLMs, including both code-specialized models and general-purpose models. Our findings indicate consistent performance improvements across all models, which underlines the practicality of CatCoder . Furthermore, we evaluate the time consumption of CatCoder in a large open source repository, and the results demonstrate the scalability of CatCoder .

Unit tests capture both functional checks and domain-specific knowledge, but this knowledge remains locked within individual projects and is rarely reused across libraries with overlapping functionality. Existing migration techniques based on structural code mappings (e.g., API signatures) often break down under divergent designs or cross-language settings, resulting in non-executable migrated tests. In this paper, we present IntentTester, a multi-agent framework for intent-driven test reuse. Instead of translating raw code, IntentTester abstracts tests into a language-agnostic Test Description Language (TDL), aligns them with semantically related entities and dependencies in a repository graph, and synthesizes executable tests through LLM-guided reasoning and iterative validation. This design enables cross-library and cross-language migration without manual intervention, producing migrated tests that existing structure-mapping approaches cannot achieve. We evaluate IntentTester on nine open-source projects across three domains (JSON, HTML, and Time) and two languages (Java and Python). IntentTester generates 2,776 syntactically correct tests with 85% correctness; in comparison, the two baselines achieve 51% and 43%. Among them, 2,410 tests executed successfully, yielding a 74% effectiveness rate. Beyond higher success rates, IntentTester also surfaced previously unknown defects including stack overflows, null dereferences, and parsing inconsistencies, several of which have been acknowledged or patched by maintainers. Our results show that intent-driven migration shifts the focus from code mappings to semantic alignment, allowing practical cross-library and cross-language test reuse while improving test quality and exposing implementation flaws.

Maintenance is a critical stage in the software lifecycle, ensuring that post-release systems remain reliable, efficient, and adaptable. However, manual software maintenance is labor-intensive, time-consuming, and error-prone, which highlights the urgent need for automation. Learning from maintenance activities conducted on other software systems offers an effective way to improve efficiency. In particular, recent research has demonstrated that migration-based approaches transfer knowledge, artifacts, or solutions from one system to another and show strong potential in tasks such as API evolution adaptation, software testing, and migrating patches for fault correction. This makes migration-based maintenance a valuable research direction for advancing automated maintenance. This paper takes a step further by presenting the first systematic research agenda on migration-based approaches to software maintenance. We characterize the migration-based maintenance lifecycle through four key stages: 182 identifying a maintenance task that can be addressed through migration, 183 selecting suitable migration sources for the target project,184 matching relevant data across systems and adapting the migrated data to the target context, and 185 validating the correctness of the migration. We also analyze the challenges that may arise at each stage. Our goal is to encourage the community to explore migration-based approaches more thoroughly and to tackle the key challenges that must be solved to advance automated software maintenance.

Large Language Models (LLMs) have demonstrated remarkable potential in automating software development tasks. While recent advances leverage Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) to align models with human preferences, the optimal training strategy remains unclear across diverse code preference types. This paper systematically investigates the roles of SFT and DPO in aligning LLMs with different code preferences. Through both theoretical analysis and empirical observation, we hypothesize that SFT excels in types with objectively verifiable optimal solutions, while applying SFT followed by DPO (S&D) enables models to explore superior solutions in types without objectively verifiable optimal solutions. Based on the analysis and experimental evidence, we propose A daptive P reference O ptimization (APO), a dynamic integration approach that adaptively amplifies preferred responses, suppresses dispreferred ones, and encourages exploration of potentially superior solutions during training. Extensive experiments across six representative code preference tasks validate our theoretical hypotheses and demonstrate that APO consistently matches or surpasses the performance of existing SFT and S&D strategies. Our work provides both theoretical foundations and practical guidance for selecting appropriate training strategies in different code preference alignment types.

Smart contracts underpin a wide range of decentralized applications—from financial services to supply-chain management—but their immutability and direct control of assets magnify the impact of any security bugs. Although many fuzz approaches have been proposed and have demonstrated their effectiveness in uncovering vulnerabilities, existing methods often rely on unguided random mutation scheduling, generate redundant inputs, and fail to adapt to smart contract-specific characteristics. To overcome these challenges, we present FuzzMaster, a feedback-driven fuzzing framework that combines deep reinforcement learning (DRL) with lightweight probabilistic scheduling to steer mutation selection at runtime intelligently. By continuously analyzing execution feedback—code coverage, function-call sequences, and vulnerability signals—FuzzMaster’s DRL agent and probabilistic tables prioritize high-impact mutations and avoid wasted effort on redundant seeds. On standard VeriSmart and SmartBugs benchmarks, FuzzMaster achieves a 66.2% detection rate with 100% precision (versus 46.9% for ItyFuzz and 43.1% for Confuzzius) and uncovers most bugs within the first second of execution. Meanwhile, in real-world Ethereum contracts, FuzzMaster identified 97 vulnerabilities in 6 categories. These results demonstrate that dynamic, vulnerability-aware mutation scheduling can dramatically improve both the efficiency and effectiveness of smart contract fuzz testing.

Deep learning libraries like Transformers and Megatron are now widely adopted in modern AI programs. However, when these libraries introduce defects, ranging from silent computation errors to subtle performance regressions, it is often challenging for downstream users to assess whether their own programs are affected. Such impact analysis requires not only understanding the defect semantics but also checking whether the client code satisfies complex triggering conditions involving configuration flags, runtime environments, and indirect API usage. We present DepRadar, an agent coordination framework for fine grained defect and impact analysis in DL library updates. DepRadar coordinates four specialized agents across three steps: 1. the PR Miner and Code Diff Analyzer extract structured defect semantics from commits or pull requests, 2. the Orchestrator Agent synthesizes these signals into a unified defect pattern with trigger conditions, and 3. the Impact Analyzer checks downstream programs to determine whether the defect can be triggered. To improve accuracy and explainability, DepRadar integrates static analysis with DL-specific domain rules for defect reasoning and client side tracing. We evaluate DepRadar on 157 PRs and 70 commits across two representative DL libraries. It achieves 90

Solidity has undergone 116 version iterations between August 2015 and February 2026, during which compiler updates have introduced behavioral changes, including issues later fixed in subsequent releases. Contracts compiled under specific versions may exhibit version-dependent execution behaviors, particularly in low-level code. These differences are often difficult for developers and users to recognize, creating opportunities for adversaries to exploit legacy compiler behaviors and deploy contracts with potentially deceptive outcomes. We define this issue as the Compiler Version Discrepancy (CVD) risk, where attackers leverage compiler-version-dependent behaviors to produce misleading or unfair outcomes while contracts appear functionally benign. We summarize five representative CVD risk patterns from real compiler inconsistencies.To mitigate this risk, we develop the CompileGuard detection tool. It combines Abstract Syntax Tree (AST) analysis, taint analysis, and symbolic execution with Control Flow Graph (CFG) analysis to identify version-sensitive code patterns. Evaluation on 227 smart contracts shows CompileGuard achieves an overall F1 score of 95.22%. A user study with 21 blockchain practitioners shows contracts exploiting CVD risks can mislead users, while detection reports enable all participants to correctly identify risk-inducing behaviors. These results highlight the practical exploitability of CVD risks and the effectiveness of automated detection in preventing such deception.

Click-through rate (CTR) prediction is a fundamental task in modern recommender systems. In recent years, the integration of large language models (LLMs) has been shown to effectively enhance the performance of traditional CTR methods. However, existing LLM-enhanced methods often require extensive processing of detailed textual descriptions for large-scale instances or user/item entities, leading to substantial computational overhead. To address this challenge, this work introduces LLaCTR, a novel and lightweight LLM-enhanced CTR method that employs a field-level enhancement paradigm. Specifically, LLaCTR first utilizes LLMs to distill crucial and lightweight semantic knowledge from small-scale feature fields through self-supervised field-feature fine-tuning. Subsequently, it leverages this field-level semantic knowledge to enhance both feature representation and feature interactions. In our experiments, we integrate LLaCTR with six representative CTR models across four datasets, demonstrating its superior performance in terms of both effectiveness and efficiency compared to existing LLM-enhanced methods. Our code is available at https://github.com/istarryn/LLaCTR.

Smart contracts, primarily written in Solidity, are Turing-complete programs on platforms like Ethereum, requiring gas fees for deployment and execution. Gas quantifies computational costs, and inefficient contracts result in unnecessary expenses for developers and users. Gas optimization at the source code level has been studied in various related works; however, existing methods for summarizing gas-inefficient patterns primarily rely on author-defined rules or heuristic approaches, and their evaluations lack a labeled dataset. In this paper, we conduct a comprehensive empirical study on the issue of gas optimization in smart contracts. We begin by gathering audit reports from Code4rena, a well-known smart contract audit platform. These reports include both expert evaluations, conducted by professionals known as Wardens, and automated analyses generated by the platform’s static analysis tool, 4naly3er. After filtering out false-positive gas optimization instances from the automated reports, we identify 2,095 instances of gas-inefficient patterns across 54 projects. We categorize these inefficiencies into 24 types using thematic analysis and find that static analysis tools often produce false positives and negatives. To address this, we propose a hybrid method combining static analysis and large language models (LLMs) to detect and repair gas inefficiencies. The static analysis tool identifies potential optimization opportunities, while the LLM refines these findings and suggests effective repairs. Our evaluation shows that our approach achieves a precision rate of 82.28% and a recall rate of 88.46%, and can save 919 units of gas per function on average during execution.

Solana is an emerging blockchain platform designed for high throughput and low transaction fees, making it inexpensive to submit transactions at scale and, consequently, increasing exposure to bot spamming and related financial exploitation. Solana bots are typically off-chain software systems that operate in a competitive on-chain execution environment by constructing and submitting transactions, and the bot-related transactions on the decentralized exchanges exceed 250 million dollars in daily trading volume in January 2026. Prior studies on Solana have examined system performance, smart-contract security, and specific on-chain phenomena. However, we still lack a systematic understanding of what Solana bots implement in practice and how these implementations manifest as observable on-chain execution fingerprints. To address this gap, we performed a large-scale empirical study of Solana bots from two complementary views: (i) 586 bot repositories collected from GitHub, and (ii) 200 bot addresses on Solana, with over 44 million on-chain transactions. Our study derives an implementation-grounded taxonomy of Solana bots comprising 15 categories grouped into five domains (e.g., Trading Operations, MEV, and On-chain Analytics), identifies a largely shared five-stage operational pipeline manifested in bot implementations, and uncovers systematic variation in on-chain trading behaviors of Solana bots across diverse trading platforms and assets. Based on our findings, we highlight future research directions, and provide recommendations for building and operating bots on the Solana blockchain.

Vulnerability exploits play a crucial role in assessing the downstream impact of Java library vulnerabilities. While some vulnerabilities are accompanied by disclosed exploit references, automatically reproducing such references into runnable exploits remains challenging because they are often incomplete, unstructured, or only describe partial reproduction steps. Recent code agents provide a promising way to automate this process, but our study shows that their generated exploits often appear successful without triggering the actual vulnerable logic, such as replacing vulnerable APIs with self-implemented functions. To address this, we propose Refploit, an LLM-based trajectory recovery framework for facilitating vulnerability reproduction from public exploit references. The key insight is that a failed agent trajectory is not entirely useless. It may have already completed some reproduction subtasks while also revealing misleading directions that should be avoided. Refploit first validates an agent-generated exploit through differential execution. When the exploit is ineffective, Refploit analyzes its reproduction progress, locates the trajectory segments associated with the reproduction progress, and derives constraints to guide focused recovery. We evaluate Refploit on three open-source Java vulnerability datasets, covering 172 exploit references for 143 vulnerabilities. Under DeepSeek-V4-Flash, Refploit successfully reproduces 138 exploits, achieving a reproduction rate of 80.2

Constructing user interfaces (UIs) is one of the most resource-intensive tasks in mobile development, often consuming more than half of overall effort. Although declarative frameworks such as Jetpack Compose (Android) and SwiftUI (iOS) have become mainstream, the majority of existing Android apps still rely on legacy XML-based layouts. Migrating these UIs to declarative paradigms is essential for maintainability and cross-platform reuse, but manual migration is costly, error-prone, and difficult to scale. We present GUIMigrator, a semantics-preserving framework that automates the migration of Android XML-based UIs to Jetpack Compose and SwiftUI. We design the Semantic UI Transpiler (SUT), which abstracts layout structures and resource semantics from legacy XML and systematically re-expresses them using the component abstractions and idioms of modern declarative frameworks. This design ensures that migrated UIs preserve both visual fidelity and functional equivalence, while generating idiomatic, compilable code that maintains cross-platform consistency with minimal manual intervention. By separating semantic interpretation from platform-specific realization, GUIMigrator provides a deterministic yet extensible basis for cross-platform modernization, avoiding the unpredictability of purely generative approaches. We evaluate GUIMigrator on 31 open-source applications across ten domains. Results show that GUIMigrator achieves high migration completeness and strong visual similarity (81.9

Repository-level code agents have shown strong promise in real-world feature addition tasks, making reliable evaluation of their capabilities increasingly important. However, existing benchmarks primarily evaluate these agents as black boxes based on final test correctness, providing limited insight into how they reason and where failures arise. To address this limitation, we introduce RACE-bench, a reasoning-augmented benchmark for evaluating code agents on repository-level feature addition tasks. RACE-bench contains 528 real-world feature addition instances from 12 open-source repositories. Each instance is paired with executable patch verification and structured intermediate reference reasoning covering issue understanding, file localization, implementation tasks, and step decomposition. Based on this design, we introduce a dual-track evaluation framework that jointly measures patch correctness and intermediate reasoning alignment with developer-accepted reference trajectories. We evaluate three representative repository-level code agents on RACE-bench. On the full benchmark, Resolved Rate ranges from 29

Large Language Models (LLMs) and pre-trained Language Models (LMs) have achieved impressive success on many software engineering tasks (e.g., code completion and code generation). By leveraging huge existing code corpora (e.g., GitHub), these models can understand the patterns in source code and use these patterns to predict code properties. However, LLMs under few-shot learning perform poorly on non-generative tasks (e.g., fault localization and vulnerability localization), and fine-tuning LLMs is time-consuming and costly for end users and small organizations. Furthermore, the performance of fine-tuning LMs for non-generative tasks is impressive, yet it heavily depends on the amount and quality of data. As a result, the current lack of data and the high cost of collecting it in real-world scenarios further limit the applicability of LMs. In this paper, we leverage the powerful generation capabilities of LLMs to enhance pre-trained LMs. Specifically, we use LLMs to generate domain-specific data, thereby improving the performance of pre-trained LMs on the target tasks. We conduct experiments by combining different LLMs in our generation phase and introducing various LMs to learn from the LLM-generated data. Then, we compare the performance of these LMs before and after learning the data. We find that LLM-generated data significantly enhances the performance of LMs. The improvement can reach up to 58.36% for fault localization and up to 6.09% for clone detection.

Open-source software supply chain security relies heavily on assessing affected versions of library vulnerabilities. While prior studies have leveraged exploits for verifying vulnerability affected versions, they point out a key limitation that exploits are version-specific and cannot be directly applied across library versions. Despite being widely acknowledged, this limitation has not been systematically validated at scale, leaving the actual applicability of exploits across versions unexplored. To fill this gap, we conduct the first large-scale empirical study on exploit applicability across library versions. We construct a comprehensive dataset consisting of 259 exploits spanning 128 Java libraries and 28,150 historical versions, covering 61 CWEs that account for 76.33

Fine-tuning is an essential and pervasive functionality for applying large language models (LLMs) to downstream tasks. However, it has the potential to substantially degrade safety alignment, e.g., by greatly increasing susceptibility to jailbreak attacks, even when the fine-tuning data is entirely harmless. Despite garnering growing attention in defense efforts during the fine-tuning stage, existing methods struggle with a persistent safety-utility dilemma: emphasizing safety compromises task performance, whereas prioritizing utility typically requires deep fine-tuning that inevitably leads to steep safety declination. In this work, we address this dilemma by shedding new light on the geometric interaction between safety- and utility-oriented gradients in safety-aligned LLMs. Through systematic empirical analysis, we uncover three key insights: (I) safety gradients lie in a low-rank subspace, while utility gradients span a broader high-dimensional space; (II) these subspaces are often negatively correlated, causing directional conflicts during fine-tuning; and (III) the dominant safety direction can be efficiently estimated from a single sample. Building upon these novel insights, we propose safety-preserving fine-tuning (SPF), a lightweight approach that explicitly removes gradient components conflicting with the low-rank safety subspace. Theoretically, we show that SPF guarantees utility convergence while bounding safety drift. Empirically, SPF consistently maintains downstream task performance and recovers nearly all pre-trained safety alignment, even under adversarial fine-tuning scenarios. Furthermore, SPF exhibits robust resistance to both deep fine-tuning and dynamic jailbreak attacks. Together, our findings provide new mechanistic understanding and practical guidance toward always-aligned LLM fine-tuning.

Personal AI agents increasingly rely on long-term memory to provide persistent personalization across sessions. However, existing memory pipelines are largely driven by semantic similarity: memory data close to the current query is retrieved and injected into the model context. This creates a critical trustworthiness gap, since a semantically related memory may still be contextually inappropriate, leading to threats such as cross-domain leakage, sycophancy, tool-call drift, or memory-induced jailbreaks. In this paper, we study memory search as a trust boundary in personal AI agents. We evaluate representative agentic memory frameworks, including A-Mem, Mem0, and MemOS, together with OpenClaw, a real-world personal-agent environment with persistent state and tool-use capability. Our results show that long-term memory is not merely a utility layer, but a durable control channel that can reshape how agents interpret tasks and execute actions, leaving them highly susceptible to the aforementioned threats. To mitigate these vulnerabilities, we propose MemGate, a lightweight and deployable memory plug-in for trustworthy memory search, with only 9M parameters and a 35.1MB footprint. MemGate is inserted between the vector memory store and the backbone LLM, requiring no LLM modification, memory-database rewriting, or inference-time LLM judge. It applies a query-conditioned neural gate to candidate memory representations, turning raw similarity search into task-conditioned memory admission. Across multiple mainstream memory frameworks, real-world agent settings, and diverse LLM backbones, MemGate reduces memory-induced threats while preserving long-term memory utility.