← back to paper
arxiv: 2605.07001 · 2 revisions
SmellBench: Evaluating LLM Agents on Architectural Code Smell Repair