Akademik Veri Yönetim Sistemi
5th International Conference on Informatics and Software Engineering, IISEC 2026, Ankara, Türkiye, 5 - 06 Şubat 2026, ss.746-751, (Tam Metin Bildiri)
Domain Model Degradation increases architectural technical debt by violating ubiquitous language and bounded contexts. This paper presents DDD-Enforcer, a VS Code extension that enforces DDD principles through real-time analysis combining Abstract Syntax Tree (AST) inspection with Large Language Models (LLMs). The system employs a four-stage multi-agent architecture to extract domain models from SRS-related documents and integrates a Retrieval-Augmented Generation (RAG) pipeline for requirement traceability. Experiments demonstrate 100% detection accuracy across 15 violation cases with an average latency of 4.49 seconds. The RAG component achieves 76.8% Top-1 accuracy, remaining robust under noisy conditions. Additionally, preliminary results for the MVC Test Orchestrator suggest potential for automated testing workflows. These findings confirm that hybrid LLM-static analysis effectively maintains DDD compliance.