Muhammad Waseem, Aakash Ahmad, Pekka Abrahamsson · 2026 · arXiv
DOI: 10.48550/arXiv.2610.24790 · open-access PDF
Große Sprachmodelle (LLMs) werden zunehmend für Refactor-Quelldateien verwendet, aber Datei-Local-Evaluationen zeigen keine Änderungen an Verträgen, Abhängigkeiten oder Anruferannahmen, die die Repository-Adoption beeinflussen. Um diese Lücke zu beheben, führen wir 1.500 Generationsversuche über 150 Produktionsdateien aus 15 Open-Source-Java-Repositorys, fünf LLM-Konfigurationen und zwei abhängigkeitsbewusste Generationsbedingungen durch: Single-Prompt-Refactoring und LangGraph-basierte Multi-Agent-Refactoring.
Machine translation (Meta M2M100); the English original follows.
Large Language Models (LLMs) are increasingly used to refactor source files, but file-local evaluations do not reveal changes to contracts, dependencies, or caller assumptions affecting repository adoption. To address this gap, we conduct 1,500 generation attempts across 150 production files from 15 open-source Java repositories, five LLM configurations, and two dependency-aware generation conditions: single-prompt refactoring and LangGraph-based multi-agent refactoring. Of these attempts, 1,455 yield complete candidates eligible for analysis. We evaluate these candidates using local changes,
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.24790, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange