AI RESEARCH
Refute-or-Promote: An Adversarial Stage-Gated Multi-Agent Review Methodology for High-Precision LLM-Assisted Defect Discovery
arXiv CS.AI
•
ArXi:2604.19049v1 Announce Type: cross LLM-assisted defect discovery has a precision crisis: plausible-but-wrong reports overwhelm maintainers and degrade credibility for real findings. We present Refute-or-Promote, an inference-time reliability pattern combining Stratified Context Hunting (SCH) for candidate generation, adversarial kill mandates, context asymmetry, and a Cross-Model Critic