Multimodal Fusion for Vulnerability Detection: Integrating Sequence and Graph-Based Analysis with LLM Augmentation

The increasing number of smart contracts has gained significant attention to the urgency of robust and scalable vulnerability detection techniques to mitigate substantial financial risks associated with their immutable nature on blockchain platforms. This paper introduces structured reasoning prompts using agent-role chaining for vulnerability detection without fine-tuning that utilizes model...