Code2UML: Agentic LLMs with context engineering for scalable software visualization

2026-05-23 · Alin-Gabriel Văduva, Anca-Ioana Andreescu, Simona-Vasilica Oprea, Adela Bâra arxiv

Large Language Model (LLM)-based code analysis tools are adopted to automate software documentation tasks. However, the scalability of these approaches to real codebases, where Intermediate Representations (IR) exceed LLM context limits, remains underexplored. This paper introduces an agentic architecture with context engineering for automated UML diagram generation from source code repositories. It employs a hierarchy of five specialized agents: PlannerAgent, AnalyzerAgent, DiagramAgent, CorrectorAgent and DependencyAnalyzerAgent, built on the Claude Agent SDK, each addressing a distinct cognitive subtask. A deterministic, importance-weighted IR compaction layer transforms full project IRs into diagram-specific views guaranteed to fit within token constraints, requiring no LLM calls and completing in milliseconds. Thus, we evaluate the system across 12 open-source repositories in 4 programming languages (Java, JavaScript, PHP, Python) and 7 UML diagram types, producing 84 observations assessed on 5 automated metrics. Results demonstrate high syntactic validity (mean: 91.5%, with component and deployment diagrams reaching 100%), strong relationship precision (mean: 0.858) and consistent structural quality (mean: 81.7/100, with cross-language variance of 3.1 points). Entity recall averaged 0.313, reflecting deliberate architectural prioritization over exhaustive coverage. A sensitivity analysis (31 to 4,578 IR entities) confirms that quality scores remain stable regardless of scale.

📄 PDF Abstract BibTeX arXiv:2605.24453

Code (0)

등록된 구현이 없습니다.

Agentic Rubrics as Contextual Verifiers for SWE Agents

2026-01-07 · Mohit Raghavendra, Anisha Gunjal, Bing Liu, Yunzhong He arxiv

Verification is critical for improving agents: it provides the reward signal for Reinforcement Learning and enables inference-time gains through Test-Time Scaling (TTS). Despite its importance, verification in software e…

Reinforcement Learning

CEDAR: Context Engineering for Agentic Data Science

2026-01-10 · Rishiraj Saha Roy, Chris Hinze, Luzian Hahn, Fabian Kuech arxiv

We demonstrate CEDAR, an application for automating data science (DS) tasks with an agentic setup. Solving DS problems with LLMs is an underexplored area that has immense market value. The challenges are manifold: task c…

Code Generation

ABC-Bench: Benchmarking Agentic Backend Coding in Real-World Development

2026-01-16 · Jie Yang, Honglin Guo, Li Ji, Jiazheng Zhou 외 arxiv

The evolution of Large Language Models (LLMs) into autonomous agents has expanded the scope of AI coding from localized code generation to complex, repository-level, and execution-driven problem solving. However, current…

Code Generation

Beyond Correctness: Enhancing Architectural Reasoning in Code LLMs via Scalable Labeling with Agentic Judgment

2026-06-12 · Kirill Vasilevski, Ximing Dong, Benjamin Rombaut, Milad Soltany 외 arxiv

LLMs have substantially improved software engineering yet real-world development requires architectural understanding. Such understanding is prohibitively expensive to label manually and impossible to verify through test…

TRAIL: Trace Reasoning and Agentic Issue Localization

2025-05-13 · Darshan Deshpande, Varun Gangal, Hersh Mehta, Jitin Krishnan 외

The increasing adoption of agentic workflows across diverse domains brings a critical need to scalably and systematically evaluate the complex traces these systems generate. Current evaluation methods depend on manual, d…

Information Retrieval