Reproduce “RefactorAssist: Agentic Refinement for Reliable Code Refactoring” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jonathan Cordeiro, Shayan Noei, Ying Zou
RefactorAssist tests AI-generated Java refactorings, applies simple static fixes, then uses compiler and test failures to guide iterative repairs by StarCoder2. The authors evaluate it on 10,000 method–test pairs from 10 open-source Java projects, counting compilation and passing associated tests as preserved behavior. Pass rates rise from 68.4% to 93.4% after ten iterations; this does not prove semantic equivalence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.