Generating Move Smart Contracts based on Concepts

Rabimba Karanjai, Sam Blackshear, Lei Xu, Weidong Shi Dec 17, 2024
arXiv PDF
BibTeX Citation 
@misc{karanjai2024conmover,
  title={Generating Move Smart Contracts based on Concepts},
  author={Karanjai, Rabimba and Blackshear, Sam and Xu, Lei and Shi, Weidong},
  howpublished={arXiv preprint arXiv:2412.12513},
  year={2024}
}
Copied to clipboard
Read Paper Download PDF

Abstract

The growing adoption of formal verification for smart contracts has spurred the development of new verifiable languages like Move. However, the limited availability of training data for these languages hinders effective code generation by large language models (LLMs). This paper presents ConMover, a novel framework that enhances LLM-based code generation for Move by leveraging a knowledge graph of Move concepts and a small set of verified code examples. ConMover integrates concept retrieval, planning, coding, and debugging agents in an iterative process to refine generated code. Evaluations with various open-source LLMs demonstrate substantial accuracy improvements over baseline models. These results underscore ConMover’s potential to address low-resource code generation challenges, bridging the gap between natural language descriptions and reliable smart contract development.

Type Preprint
Venue arXiv preprint arXiv:2412.12513
Date Dec 2024
Smart Contracts Blockchain Code Generation LLMs First Author
Smart Contracts Blockchain Code Generation LLMs First Author