RAG-Enhanced LLMs for Automated Software Testing and Code Inspection Reduce Hallucination in V&V Pipelines

arXiv·low signal

This paper implements retrieval-augmented generation to address LLM hallucination in automated test case generation and source code inspection for software verification and validation. By grounding LLM outputs in retrieved documentation and code context, the approach reduces confidently incorrect outputs — a critical failure mode when LLMs are used in formal SDLC processes where incorrect test cases or missed defects have direct quality impact.