Assessing GitHub Copilot in Solidity Development: Capabilities, Testing, and Bug Fixing

Abstract

In the rapidly evolving landscape of blockchain technology, the development of reliable and secure smart contracts represents one of several crucial challenges. GitHub Copilot, an AI-powered code assistant, aims to enhance developer productivity by generating code snippets, facilitating testing, and assisting in program repair. This research examines Copilot’s proficiency in generating functional and secure smart contracts, including token creation adhering to standards such as ERC20, ERC721, and ERC1155 with various optional features. Additionally, the study assesses its effectiveness in common development tasks, including the implementation of widely employed libraries such as SafeMath. Through controlled experiments, the accuracy, efficiency, and security of the code generated by Copilot are evaluated. This evaluation identifies both its strengths in expediting the development process and its limitations in managing complex blockchain-specific logic and security considerations. The findings contribute to an expanded understanding of the role of AI-assisted programming in blockchain development, offering insights into how developers can best leverage such tools in creating and testing smart contracts. This research aims to guide both practitioners and researchers in the blockchain domain, advancing the discussion on integrating AI into software development workflows in the context of Solidity and smart contract development, underscoring the need for further research to address the challenges and opportunities presented by AI in blockchain technology.

Keywords

• GitHub copilot
• solidity
• smart contract
• AI
• bug fixing
• vulnerability