Agentic Specification Generator for Move Programs

2025-09-29

AI Summary

Emerging verification-oriented programming languages like Move lack LLM-driven tools for automated specification generation. Method: This paper introduces the first automated specification generation framework tailored for Move smart contracts. It adopts an agent-based modular architecture that deeply integrates Move’s language features with formal specification modeling and incorporates a verification-toolchain feedback loop for iterative output refinement. Contribution/Results: Compared to conventional approaches, the framework significantly enhances code understanding and specification generation for non-mainstream languages. Experiments show it generates verifiable specifications for 84% of test functions, increasing the number of verifiable clauses by 57% over baselines; integrating feedback further improves specification quality by 30%. This work represents the first systematic exploration of large language models’ adaptation pathways and potential within verification-first language ecosystems, establishing a novel paradigm for LLM-augmented formal methods.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

While LLM-based specification generation is gaining traction, existing tools primarily focus on mainstream programming languages like C, Java, and even Solidity, leaving emerging and yet verification-oriented languages like Move underexplored. In this paper, we introduce MSG, an automated specification generation tool designed for Move smart contracts. MSG aims to highlight key insights that uniquely present when applying LLM-based specification generation to a new ecosystem. Specifically, MSG demonstrates that LLMs exhibit robust code comprehension and generation capabilities even for non-mainstream languages. MSG successfully generates verifiable specifications for 84% of tested Move functions and even identifies clauses previously overlooked by experts. Additionally, MSG shows that explicitly leveraging specification language features through an agentic, modular design improves specification quality substantially (generating 57% more verifiable clauses than conventional designs). Incorporating feedback from the verification toolchain further enhances the effectiveness of MSG, leading to a 30% increase in generated verifiable specifications.

Problem

• Automates specification generation for Move smart contracts using LLMs
• Addresses the gap in specification tools for verification-oriented Move language
• Improves specification quality through agentic design and verification feedback

Innovation

• Automated specification generation for Move smart contracts
• Agentic modular design leveraging specification language features
• Incorporating verification toolchain feedback to enhance effectiveness