Reproduce “Escaping the Self-Repair Trap: Improving Test Oracle Generation via Dual-Context Awareness” 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
Kefan Li, Hongyue Yu, Yuan Yuan
DCAware completes one missing assertion in an existing regression test by combining relevant code context with selected values queried from a running test. It writes the assertion without an execution-and-repair loop. On 500 Java test instances, GPT-5-mini achieved an 88.20% Pass Rate, versus 73.20% for ChatAssert; on 183 shared passed instances, its Shared Kill Golden score was 92.83%, versus 86.31%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.