AxDafny: Agentic Verified Code Generation in Dafny

📰 ArXiv cs.AI

Learn how AxDafny generates verified code in Dafny using agentic code generation and verification techniques, and how to apply these methods to improve code reliability

advanced Published 1 Jul 2026

Action Steps

Implement AxDafny to generate verified code in Dafny
Use LiveCodeBench-Pro-Dafny to test and evaluate the performance of AxDafny
Apply verifier-guided repair techniques to improve code generation
Integrate AxDafny with existing development workflows to enhance code reliability
Evaluate the effectiveness of AxDafny in reducing errors and improving code quality

Who Needs to Know This

Software engineers and researchers working on formal verification and code generation can benefit from this article, as it provides a new approach to generating reliable code

Key Insight

💡 AxDafny's verifier-guided repair framework can generate both executable code and proof artifacts for verification, improving code reliability

Key Takeaways

Learn how AxDafny generates verified code in Dafny using agentic code generation and verification techniques, and how to apply these methods to improve code reliability

Full Article

Title: AxDafny: Agentic Verified Code Generation in Dafny

Abstract:
arXiv:2606.32007v1 Announce Type: new Abstract: We study agentic code generation in Dafny, where a model must generate both executable code and the proof artifacts for verification. We present AxDafny, a verifier-guided repair framework that iteratively generates implementations, invariants, assertions, and termination arguments. We also introduce LiveCodeBench-Pro-Dafny (LCB-Pro-Dafny), a benchmark of 250 competition-style programming problems translated into Dafny with formal specifications an

Read full paper → ← Back to Reads