The findings demonstrate that structural evaluations of LLM-generated decompositions should explicitly control for mapping coverage, as apparent differences between prompting strategies may otherwise reflect methodological bias rather than genuine architectural quality.
Abstract
Decomposing monolithic systems into microservices is a key activity in software modernization. Although Large Language Models (LLMs) can generate semantically plausible decompositions from textual requirements, it remains unclear whether these proposals preserve the structural dependencies implemented in the source code. This paper evaluates the structural adherence of microservice decompositions generated by OpenAI o3 for the PetClinic and Bookstore systems. We propose an automated validation pipeline based on static dependency analysis and compare zero-shot and few-shot prompting using dependency preservation (TPD) and dependency violation (TVD) metrics. A robustness analysis was conducted to control for differences in class-to-service mapping coverage. After normalization, both prompting strategies produced equivalent structural adherence, achieving TPD values of 68.0% (PetClinic) and 83.3% (Bookstore). The findings demonstrate that structural evaluations of LLM-generated decompositions should explicitly control for mapping coverage, as apparent differences between prompting strategies may otherwise reflect methodological bias rather than genuine architectural quality.
Microservice architectures have become dominant for modernizing monolithic systems, yet identifying appropriate services remains challenging and largely manual. Existing decomposition approaches are predominantly code-centric, limiting applicability in early design stages where only textual requirements are available. Despite advances in Large Language Models (LLMs), limited empirical evidence exists on their ability to synthesize complete microservice architectures from natural-language requirements, including service definitions and inter-service interactions. This study investigates whether an LLM can bridge requirements engineering and architectural design, generating architectures solely from textual requirements and evaluating structural agreement and perceived quality of results. We conduct a mixed-method study using OpenAI o3 under zero-shot (ZS) and few-shot (FS) prompting across two systems (Bookstore, PetClinic), one execution per system/condition. Architectures are evaluated through (i) comparison with reference architectures using precision, recall, and F1-score for service identification and communication recovery, and (ii) a blinded expert assessment of correctness, completeness, modularity, and plausibility, plus open feedback synthesis. OpenAI o3 identifies services with higher agreement under FS prompting (F1 = 0.79 for ZS versus = 0.97 for FS). Communication recovery is more challenging: ZS produces dense architectures with high recall but low precision (F1 = 0.61), while FS improves agreement, reaching F1 = 0.82 and reducing unsupported dependencies. Expert evaluation corroborates these results, with FS architectures perceived as more modular, coherent, and plausible than ZS outputs. OpenAI o3 shows potential for requirements-driven synthesis when guided by exemplar prompting. Results are model- and context-specific from two small systems, not model-independent proof.
D. Albuquerque, José Renan, Guillermo Rodríguez et al.· 0 citations
Microservice-based systems evolve continuously through API changes, service splits and merges, dependency churn, and deployment-topology drift driven by DevOps practices. This rapid evolution erodes the assumptions encoded in formal models, test suites, and architectural descriptions, leaving practitioners with limited guidance on how to keep verification artifacts aligned with evolving implementations. Prior surveys cover microservices design, deployment, performance, and isolated verification techniques, but, to our knowledge, none consolidate the intersection of evolution and formal/rigorous verification, nor map how repository-derived signals and CI/CD pipelines support continuous verification of microservice-based systems. This paper addresses that gap through a systematic literature review (SLR). Following the Kitchenham et al. guidelines and PRISMA-style reporting, we defined a review protocol with four research questions covering (i) formal modeling and verification approaches, (ii) the impact of architectural and API evolution on verifiability, (iii) repository-derived signals for updating formal artifacts, and (iv) continuous and incremental verification in CI/CD pipelines. Structured searches in IEEE Xplore, ACM Digital Library, SpringerLink, and Scopus returned 1187 records, which were screened in seven stages and reduced to 18 included studies (13 primary studies that supply evidence and five background studies retained for transparency and framing) through inclusion/exclusion criteria and a weighted 15-item quality-assessment instrument. We synthesized the included studies thematically and compared them along five dimensions: verification technique, evolution dimension, automation level, empirical evidence, and tool/artifact availability. The synthesis shows that actor-based model checking, TLA+ control-plane verification, static code analysis, continuous certification, and runtime/self-adaptive testing form five complementary clusters, but empirical evaluation is dominated by small academic case studies, only a minority of approaches automate model updates from repository signals, and few are integrated end-to-end into CI/CD. We discuss methodological limitations of the body of work, including a lack of industrial-scale validation and shared datasets, contrast intrinsic (build-time, model-driven) and extrinsic (recovery-based) verification strategies, and outline a research agenda toward repository-aware, pipeline-integrated verification for evolving microservices. The present synthesis distils a set of immediate, evidence-based actions for practitioners together with a focused agenda of open research challenges.
R. Gómez, Ebeid Elsayed, Enrique R. Zarate et al.· Software· 0 citations
E EduPluginBench is introduced, an executable benchmark and staged admission method for generated plugins in governed software ecosystems that retains protocols, public-source provenance, raw generations, row-level decisions, audits, analysis code, and reproduction instructions.
The integration of Large Language Models (LLMs) into software engineering has shifted the focus from function-level generation to repository-scale assistance. However, existing benchmarks largely rely on bug reports from GitHub Issues, which often allow models to bypass genuine understanding via pattern matching on error logs. This misalignment under-measures Edit Bias, which refers to premature generation, where models prematurely propose code modifications instead of understanding the existing repository architecture. Furthermore, current LLM-as-a-Judge scalar scoring suffers from high variance and low interpretability. This work introduces RepoProbe, a novel benchmark for evaluating repository-level code understanding through open-ended Q&A using GitHub Discussions, which focuses on open-ended architectural inquiries rather than defect reporting. To ensure rigorous evaluation, we propose a Checklist-Based Verification Protocol that decomposes answers into atomic, verifiable facts, thereby replacing subjective ratings with objective verification. Our evaluation of state-of-the-art (SOTA) LLMs reveals a persistent gap between high clarity and evidencegrounded technical correctness. It also quantitatively confirms the prevalence of edit bias, in which models prioritize code generation instead of architectural analysis. Finally, we demonstrate that our verification protocol significantly improves evaluation reliability compared to traditional evaluations with scalar scoring.
Adopting Continuous Integration (CI) often requires writing YAML configurations that are error-prone and challenging to maintain. Despite increasing LLM use in software engineering, their ability to generate CI configurations from natural language across services and model families remains unclear. This paper presents a large empirical study on using LLMs to generate CI configurations. We introduce DOC2CI, a benchmark of 3,363 description-to-YAML pairs collected from the official documentation of four CI services, and evaluate 14 open-weight models from 7B-34B parameters together with GPT-4o and GPT-4.1, producing over 53,000 configurations. We assess both reference alignment and schema validity to determine whether the generated configurations are structurally valid. We further develop a failure taxonomy from a manual analysis of 385 configurations and examine why LLMs disagree. Across models and services, exact reference reproduction never exceeds 3.1%, and while 97% of outputs parse as YAML, only 71% satisfy service schemas. Larger models improve structural validity, but code specialization provides no consistent advantage over comparable general models. Model differences are driven largely by output completeness: for the same request, some models generate the expected fragment while others produce a full workflow. Finally, a training-free schema-guided repair method improves schema validity to 94%, while fine-tuning improves similarity to documentation but reduces standalone validity. This suggests that similarity and validity are distinct objectives for CI generation and motivate schema-aware evaluation and tooling for LLM-based configuration generation.
This paper identifies patch verbosity as a major yet overlooked concern in LLM-based APR and proposes RECAP, a lightweight, plug-and-play adapter that attaches to existing repair frameworks after generation that achieves a substantially better size-correctness tradeoff.
Wenqiang Luo, J. Keung, Xiaoyu Shi et al.· 0 citations