Integrating Model-Driven Engineering and Large Language Models for Test Scenario Generation for Smart Contracts

Large Language Models (LLMs) have demonstrated significant potential in transforming software testing by automating tasks such as test case generation. In this work, we explore the integration of LLMs within a Model-Driven Engineering (MDE) approach to enhance the automation of test case generation for smart contracts. Our focus lies in the use of Role-Based Access Control (RBAC) models as formal specifications that guide the generation of test scenarios. By leveraging LLMs’ ability to interpret both natural language and model artifacts, we enable the derivation of model-based test cases that align with specified access control policies. These test cases are subsequently translated into executable code in Digital Asset Modeling Language (DAML) targeting blockchain-based smart contract platforms. Building on prior research that established a complete MDE pipeline for DAML smart contract development, we extend the framework with LLM-supported test automation capabilities and implement the necessary tooling to support this integration. Our evaluation demonstrates the feasibility of using LLMs in this context, highlighting their potential to improve testing coverage, reduce manual effort, and ensure conformance with access control specifications in smart contract systems.