The rise of large model-based AI agents has spurred interest in Multi-Agent Systems (MAS) for their capabilities in decision-making, collaboration, and adaptability. While the Model Context Protocol (MCP) addresses tool invocation and data exchange challenges via a unified protocol, it lacks support for organizing agent-level collaboration. To bridge this gap, we propose Agent-as-a-Service based on Agent Network (AaaS-AN), a service-oriented paradigm grounded in the Role-Goal-Process-Service (RGPS) standard. AaaS-AN unifies the entire agent lifecycle, including construction, integration, interoperability, and networked collaboration, through two core components: (1) a dynamic Agent Network, which models agents and agent groups as vertexes that self-organize within the network based on task and role dependencies; (2) service-oriented agents, incorporating service discovery, registration, and interoperability protocols. These are orchestrated by a Service Scheduler, which leverages an Execution Graph to enable distributed coordination, context tracking, and runtime task management. We validate AaaS-AN on mathematical reasoning and application-level code generation tasks, which outperforms state-of-the-art baselines. Notably, we constructed a MAS based on AaaS-AN containing agent groups, Robotic Process Automation (RPA) workflows, and MCP servers over 100 agent services. We also release a dataset containing 10,000 long-horizon multi-agent workflows to facilitate future research on long-chain collaboration in MAS.
Large Language Models (LLMs) have been extensively used across diverse domains, including virtual assistants, automated code generation, and scientific research. However, they remain vulnerable to jailbreak attacks, which manipulate the models into generating harmful responses despite safety alignment. Recent studies have shown that current safety-aligned LLMs often undergo the shallow safety alignment, where the first few tokens largely determine whether the response will be harmful. Through comprehensive observations, we find that safety-aligned LLMs and various defense strategies generate highly similar initial tokens in their refusal responses, which we define as safety trigger tokens. Building on this insight, we propose \texttt{D-STT}, a simple yet effective defense algorithm that identifies and explicitly decodes safety trigger tokens of the given safety-aligned LLM to trigger the model's learned safety patterns. In this process, the safety trigger is constrained to a single token, which effectively preserves model usability by introducing minimum intervention in the decoding process. Extensive experiments across diverse jailbreak attacks and benign prompts demonstrate that \ours significantly reduces output harmfulness while preserving model usability and incurring negligible response time overhead, outperforming ten baseline methods.
The application of large language models (LLMs) in the field of coding is evolving rapidly: from code assistants, to autonomous coding agents, and then to generating complete projects through natural language. Early LLM code benchmarks primarily focused on code generation accuracy, but these benchmarks have gradually become saturated. Benchmark saturation weakens their guiding role for LLMs. For example, HumanEval Pass@1 has reached 99.4% and MBPP 94.2%. Among various attempts to address benchmark saturation, approaches based on software engineering have stood out, but the saturation of existing software engineering benchmarks is rapidly increasing. To address this, we propose a new benchmark, Web-Bench, which contains 50 projects, each consisting of 20 tasks with sequential dependencies. The tasks implement project features in sequence, simulating real-world human development workflows. When designing Web-Bench, we aim to cover the foundational elements of Web development: Web Standards and Web Frameworks. Given the scale and complexity of these projects, which were designed by engineers with 5 to 10 years of experience, each presents a significant challenge. On average, a single project takes 4 to 8 hours for a senior engineer to complete. On our given benchmark agent (Web-Agent), SOTA (Claude 3.7 Sonnet) achieves only 25.1% Pass@1, significantly lower (better) than SWE-Bench's Verified (65.4%) and Full (33.8%) scores. Finally, we discuss that in any development field, Standards and Frameworks represent foundational knowledge and efficiency tools, respectively, and LLMs require optimization tailored to them.
Large Language Models (LLMs) have demonstrated unprecedented capability in code generation. However, LLM-generated code is still plagued with a wide range of functional errors, especially for complex programming tasks that LLMs have not seen before. Recent studies have shown that developers often struggle with inspecting and fixing incorrect code generated by LLMs, diminishing their productivity and trust in LLM-based code generation. Inspired by the mutual grounding theory in communication, we propose an interactive approach that leverages code comments as a medium for developers and LLMs to establish a shared understanding. Our approach facilitates iterative grounding by interleaving code generation, inline comment generation, and contextualized user feedback through editable comments to align generated code with developer intent. We evaluated our approach on two popular benchmarks and demonstrated that our approach significantly improved multiple state-of-the-art LLMs, e.g., 17.1% pass@1 improvement for code-davinci-002 on HumanEval. Furthermore, we conducted a user study with 12 participants in comparison to two baselines: (1) interacting with GitHub Copilot, and (2) interacting with a multi-step code generation paradigm called Multi-Turn Program Synthesis. Participants completed the given programming tasks 16.7% faster and with 10.5% improvement in task success rate when using our approach. Both results show that interactively refining code comments enables the collaborative establishment of mutual grounding, leading to more accurate code generation and higher developer confidence.
Scientific diagrams are vital tools for communicating structured knowledge across disciplines. However, they are often published as static raster images, losing symbolic semantics and limiting reuse. While Multimodal Large Language Models (MLLMs) offer a pathway to bridging vision and structure, existing methods lack semantic control and structural interpretability, especially on complex diagrams. We propose Draw with Thought (DwT), a training-free framework that guides MLLMs to reconstruct diagrams into editable mxGraph XML code through cognitively-grounded Chain-of-Thought reasoning. DwT enables interpretable and controllable outputs without model fine-tuning by dividing the task into two stages: Coarse-to-Fine Planning, which handles perceptual structuring and semantic specification, and Structure-Aware Code Generation, enhanced by format-guided refinement. To support evaluation, we release Plot2XML, a benchmark of 247 real-world scientific diagrams with gold-standard XML annotations. Extensive experiments across eight MLLMs show that our approach yields high-fidelity, semantically aligned, and structurally valid reconstructions, with human evaluations confirming strong alignment in both accuracy and visual aesthetics, offering a scalable solution for converting static visuals into executable representations and advancing machine understanding of scientific graphics.
With the rise of AI-based code generation, customizing existing code out of natural language instructions to modify visual results -such as figures or images -has become possible, promising to reduce the need for deep programming expertise. However, even experienced developers can struggle with this task, as it requires identifying relevant code regions (feature location), generating valid code variants, and ensuring the modifications reliably align with user intent. In this paper, we introduce vTikZ, the first benchmark designed to evaluate the ability of Large Language Models (LLMs) to customize code while preserving coherent visual outcomes. Our benchmark consists of carefully curated vTikZ editing scenarios, parameterized ground truths, and a reviewing tool that leverages visual feedback to assess correctness. Empirical evaluation with stateof-the-art LLMs shows that existing solutions struggle to reliably modify code in alignment with visual intent, highlighting a gap in current AI-assisted code editing approaches. We argue that vTikZ opens new research directions for integrating LLMs with visual feedback mechanisms to improve code customization tasks in various domains beyond TikZ, including image processing, art creation, Web design, and 3D modeling.
Large Language Models (LLMs) have achieved remarkable success in code generation tasks, powering various applications like code completion, debugging, and programming assistance. However, existing benchmarks such as HumanEval, MBPP, and BigCodeBench primarily evaluate LLMs on English-only prompts, overlooking the real-world scenario where multilingual developers often use code-mixed language while interacting with LLMs. To address this gap, we introduce CodeMixBench, a novel benchmark designed to evaluate the robustness of LLMs on code generation from code-mixed prompts. Built upon BigCodeBench, CodeMixBench introduces controlled code-mixing (CMD) into the natural language parts of prompts across three language pairs: Hinglish (Hindi-English), Spanish-English, and Chinese Pinyin-English. We comprehensively evaluate a diverse set of open-source code generation models ranging from 1.5B to 15B parameters. Our results show that code-mixed prompts consistently degrade Pass@1 performance compared to their English-only counterparts, with performance drops increasing under higher CMD levels for smaller models. CodeMixBench provides a realistic evaluation framework for studying multilingual code generation and highlights new challenges and directions for building robust code generation models that generalize well across diverse linguistic settings.
The emergence of Generative AI (Gen AI) and Large Language Models (LLMs) has enabled more advanced chatbots capable of human-like interactions. However, these conversational agents introduce a broader set of operational risks that extend beyond traditional cybersecurity considerations. In this work, we propose a novel, instrumented risk-assessment metric that simultaneously evaluates potential threats to three key stakeholders: the service-providing organization, end users, and third parties. Our approach incorporates the technical complexity required to induce erroneous behaviors in the chatbot--ranging from non-induced failures to advanced prompt-injection attacks--as well as contextual factors such as the target industry, user age range, and vulnerability severity. To validate our metric, we leverage Garak, an open-source framework for LLM vulnerability testing. We further enhance Garak to capture a variety of threat vectors (e.g., misinformation, code hallucinations, social engineering, and malicious code generation). Our methodology is demonstrated in a scenario involving chatbots that employ retrieval-augmented generation (RAG), showing how the aggregated risk scores guide both short-term mitigation and longer-term improvements in model design and deployment. The results underscore the importance of multi-dimensional risk assessments in operationalizing secure, reliable AI-driven conversational systems.
Despite the rapid growth of machine learning research, corresponding code implementations are often unavailable, making it slow and labor-intensive for researchers to reproduce results and build upon prior work. In the meantime, recent Large Language Models (LLMs) excel at understanding scientific documents and generating high-quality code. Inspired by this, we introduce PaperCoder, a multi-agent LLM framework that transforms machine learning papers into functional code repositories. PaperCoder operates in three stages: planning, where it constructs a high-level roadmap, designs the system architecture with diagrams, identifies file dependencies, and generates configuration files; analysis, which focuses on interpreting implementation-specific details; and generation, where modular, dependency-aware code is produced. Moreover, each phase is instantiated through a set of specialized agents designed to collaborate effectively across the pipeline. We then evaluate PaperCoder on generating code implementations from machine learning papers based on both model-based and human evaluations, particularly from the authors of those papers, with author-released repositories as ground truth if available. Our results demonstrate the effectiveness of PaperCoder in creating high-quality, faithful implementations. Furthermore, it consistently shows strengths in the recently released PaperBench benchmark, surpassing strong baselines by substantial margins. Code is available at: https://github.com/going-doer/Paper2Code.
Large language models (LLMs) have transformed software development through code generation capabilities, yet their effectiveness for high-performance computing (HPC) remains limited. HPC code requires specialized optimizations for parallelism, memory efficiency, and architecture-specific considerations that general-purpose LLMs often overlook. We present MARCO (Multi-Agent Reactive Code Optimizer), a novel framework that enhances LLM-generated code for HPC through a specialized multi-agent architecture. MARCO employs separate agents for code generation and performance evaluation, connected by a feedback loop that progressively refines optimizations. A key innovation is MARCO's web-search component that retrieves real-time optimization techniques from recent conference proceedings and research publications, bridging the knowledge gap in pre-trained LLMs. Our extensive evaluation on the LeetCode 75 problem set demonstrates that MARCO achieves a 14.6\% average runtime reduction compared to Claude 3.5 Sonnet alone, while the integration of the web-search component yields a 30.9\% performance improvement over the base MARCO system. These results highlight the potential of multi-agent systems to address the specialized requirements of high-performance code generation, offering a cost-effective alternative to domain-specific model fine-tuning.
Large Language Models demonstrate the ability to solve various programming tasks, including code generation. Typically, the performance of LLMs is measured on benchmarks with small or medium-sized context windows of thousands of lines of code. At the same time, in real-world software projects, repositories can span up to millions of LoC. This paper closes this gap by contributing to the long context code generation benchmark (YABLoCo). The benchmark featured a test set of 215 functions selected from four large repositories with thousands of functions. The dataset contained metadata of functions, contexts of the functions with different levels of dependencies, docstrings, functions bodies, and call graphs for each repository. This paper presents three key aspects of the contribution. First, the benchmark aims at function body generation in large repositories in C and C++, two languages not covered by previous benchmarks. Second, the benchmark contains large repositories from 200K to 2,000K LoC. Third, we contribute a scalable evaluation pipeline for efficient computing of the target metrics and a tool for visual analysis of generated code. Overall, these three aspects allow for evaluating code generation in large repositories in C and C++.
Creative coding platforms like Scratch have democratized programming for children, yet translating imaginative ideas into functional code remains a significant hurdle for many young learners. While AI copilots assist adult programmers, few tools target children in block-based environments. Building on prior research \cite{druga_how_2021,druga2023ai, druga2023scratch}, we present Cognimates Scratch Copilot: an AI-powered assistant integrated into a Scratch-like environment, providing real-time support for ideation, code generation, debugging, and asset creation. This paper details the system architecture and findings from an exploratory qualitative evaluation with 18 international children (ages 7--12). Our analysis reveals how the AI Copilot supported key creative coding processes, particularly aiding ideation and debugging. Crucially, it also highlights how children actively negotiated the use of AI, demonstrating strong agency by adapting or rejecting suggestions to maintain creative control. Interactions surfaced design tensions between providing helpful scaffolding and fostering independent problem-solving, as well as learning opportunities arising from navigating AI limitations and errors. Findings indicate Cognimates Scratch Copilot's potential to enhance creative self-efficacy and engagement. Based on these insights, we propose initial design guidelines for AI coding assistants that prioritize youth agency and critical interaction alongside supportive scaffolding.
Teaching large language models (LLMs) to critique and refine their outputs is crucial for building systems that can iteratively improve, yet it is fundamentally limited by the ability to provide accurate judgments and actionable suggestions. In this work, we study LLM critics for code generation and propose $\texttt{CTRL}$, a framework for $\texttt{C}$ritic $\texttt{T}$raining via $\texttt{R}$einforcement $\texttt{L}$earning, which trains a critic model to generate feedback that maximizes correction performance for a fixed generator model without human supervision. Our results demonstrate that critics trained with $\texttt{CTRL}$ significantly enhance pass rates and mitigate compounding errors across both base and stronger generator models. Furthermore, we show that these critic models act as accurate generative reward models and enable test-time scaling through iterative critique-revision, achieving up to 106.1% relative improvements across challenging code generation benchmarks.
The prevalence of Large Language Models (LLMs) is revolutionizing the process of writing code. General and code LLMs have shown impressive performance in generating standalone functions and code-completion tasks with one-shot queries. However, the ability to solve comprehensive programming tasks with recursive requests and bug fixes remains questionable. In this paper, we propose EduBot, an intelligent automated assistant system that combines conceptual knowledge teaching, end-to-end code development, personalized programming through recursive prompt-driven methods, and debugging with limited human interventions powered by LLMs. We show that EduBot can solve complicated programming tasks consisting of sub-tasks with increasing difficulties ranging from conceptual to coding questions by recursive automatic prompt-driven systems without finetuning on LLMs themselves. To further evaluate EduBot's performance, we design and conduct a benchmark suite consisting of 20 scenarios in algorithms, machine learning, and real-world problems. The result shows that EduBot can complete most scenarios in less than 20 minutes. Based on the benchmark suites, we perform a comparative study to take different LLMs as the backbone and to verify EduBot's compatibility and robustness across LLMs with varying capabilities. We believe that EduBot is an exploratory approach to explore the potential of pre-trained LLMs in multi-step reasoning and code generation for solving personalized assignments with knowledge learning and code generation.
AI code generation tools have gained significant popularity among developers, who use them to assist in software development due to their capability to generate code. Existing studies mainly explored the quality, e.g., correctness and security, of AI-generated code, while in real-world software development, the prerequisite is to distinguish AI-generated code from human-written code, which emphasizes the need to explicitly declare AI-generated code by developers. To this end, this study intends to understand the ways developers use to self-declare AI-generated code and explore the reasons why developers choose to self-declare or not. We conducted a mixed-methods study consisting of two phases. In the first phase, we mined GitHub repositories and collected 613 instances of AI-generated code snippets. In the second phase, we conducted a follow-up industrial survey, which received 111 valid responses. Our research revealed the practices followed by developers to self-declare AI-generated code. Most practitioners (76.6%) always or sometimes self-declare AI-generated code. In contrast, other practitioners (23.4%) noted that they never self-declare AI-generated code. The reasons for self-declaring AI-generated code include the need to track and monitor the code for future review and debugging, and ethical considerations. The reasons for not self-declaring AI-generated code include extensive modifications to AI-generated code and the developers' perception that self-declaration is an unnecessary activity. We finally provided guidelines for practitioners to self-declare AI-generated code, addressing ethical and code quality concerns.
Large language models (LLMs) have demonstrated remarkable capabilities in code generation tasks. However, a significant gap remains between their current performance and that of expert software engineers. A key differentiator is that human engineers actively seek clarification when faced with ambiguous requirements, while LLMs typically generate code regardless of uncertainties in the problem description. We present ClarifyCoder, a novel framework with synthetic data generation and instruction-tuning that enables LLMs to identify ambiguities and request clarification before proceeding with code generation. While recent work has focused on LLM-based agents for iterative code generation, we argue that the fundamental ability to recognize and query ambiguous requirements should be intrinsic to the models themselves. Our approach consists of two main components: (1) a data synthesis technique that augments existing programming datasets with scenarios requiring clarification to generate clarification-aware training data, and (2) a fine-tuning strategy that teaches models to prioritize seeking clarification over immediate code generation when faced with incomplete or ambiguous requirements. We further provide an empirical analysis of integrating ClarifyCoder with standard fine-tuning for a joint optimization of both clarify-awareness and coding ability. Experimental results demonstrate that ClarifyCoder significantly improves the communication capabilities of Code LLMs through meaningful clarification dialogues while maintaining code generation capabilities.
Recent advancements in large language models (LLMs) have demonstrated promising capabilities in code generation tasks. However, most existing benchmarks focus on isolated functions and fail to capture the complexity of real-world, class-level software structures. To address this gap, we introduce a large-scale, Python class-level dataset curated from $13{,}174$ real-world open-source projects. The dataset contains over 842,000 class skeletons, each including class and method signatures, along with associated docstrings when available. We preserve structural and contextual dependencies critical to realistic software development scenarios and enrich the dataset with static code metrics to support downstream analysis. To evaluate the usefulness of this dataset, we use extracted class skeletons as prompts for GPT-4 to generate full class implementations. Results show that the LLM-generated classes exhibit strong lexical and structural similarity to human-written counterparts, with average ROUGE@L, BLEU, and TSED scores of 0.80, 0.59, and 0.73, respectively. These findings confirm that well-structured prompts derived from real-world class skeletons significantly enhance LLM performance in class-level code generation. This dataset offers a valuable resource for benchmarking, training, and improving LLMs in realistic software engineering contexts.
As LLM-generated code grows in popularity, more evaluation is needed to assess the risks of using such tools, especially for safety-critical applications such as path planning. Existing coding benchmarks are insufficient as they do not reflect the context and complexity of safety-critical applications. To this end, we assessed six LLMs' abilities to generate the code for three different path-planning algorithms and tested them on three maps of various difficulties. Our results suggest that LLM-generated code presents serious hazards for path planning applications and should not be applied in safety-critical contexts without rigorous testing.
Recent technical breakthroughs in large language models (LLMs) have enabled them to fluently generate source code. Software developers often leverage both general-purpose and code-specialized LLMs to revise existing code or even generate a whole function from scratch. These capabilities are also beneficial in no-code or low-code contexts, in which one can write programs without a technical background. However, due to their internal design, LLMs are prone to generating hallucinations, which are incorrect, nonsensical, and not justifiable information but difficult to identify its presence. This problem also occurs when generating source code. Once hallucinated code is produced, it is often challenging for users to identify and fix it, especially when such hallucinations can be identified under specific execution paths. As a result, the hallucinated code may remain unnoticed within the codebase. This survey investigates recent studies and techniques relevant to hallucinations generated by CodeLLMs. We categorize the types of hallucinations in the code generated by CodeLLMs, review existing benchmarks and mitigation strategies, and identify open challenges. Based on these findings, this survey outlines further research directions in the detection and removal of hallucinations produced by CodeLLMs.
Large language models (LLMs) play a crucial role in software engineering, excelling in tasks like code generation and maintenance. However, existing benchmarks are often narrow in scope, focusing on a specific task and lack a comprehensive evaluation framework that reflects real-world applications. To address these gaps, we introduce CoCo-Bench (Comprehensive Code Benchmark), designed to evaluate LLMs across four critical dimensions: code understanding, code generation, code modification, and code review. These dimensions capture essential developer needs, ensuring a more systematic and representative evaluation. CoCo-Bench includes multiple programming languages and varying task difficulties, with rigorous manual review to ensure data quality and accuracy. Empirical results show that CoCo-Bench aligns with existing benchmarks while uncovering significant variations in model performance, effectively highlighting strengths and weaknesses. By offering a holistic and objective evaluation, CoCo-Bench provides valuable insights to guide future research and technological advancements in code-oriented LLMs, establishing a reliable benchmark for the field.