Advanced Detection using LLM

Our project seeks to enhance the security of Uniswap V4 Hook Contracts by developing an advanced static analysis solution leveraging LLMs. Since Hook Contracts are a type of Smart Contract, developers can freely include custom functions and assembly code written in Yul, which adds to code complexity and makes it challenging for conventional static analysis tools to detect specific patterns or behaviors effectively.

We are exploring using LLMs to enhance static analysis and address these challenges. By training an LLM on smart contract code patterns, we aim to achieve a deep understanding of the complex structures within Hook functions and assembly code. This approach allows the LLM to recognize unique characteristics and potential vulnerabilities within Hook Contracts, enabling proactive identification of unintended behaviors or security flaws in the code. Such advanced analysis goes beyond merely detecting syntactic errors. It will offer a practical evaluation of hidden risks within Hook Contracts, contributing to a safer DeFi ecosystem on UniChain.