Search papers, labs, and topics across Lattice.
This paper introduces Lean4Agent, a novel framework that employs Lean4, a dependent-type formal language, to model and verify the workflows of large language model (LLM) agents. By creating the FormalAgentLib library, the authors enable the formal specification and debugging of agent behaviors, demonstrating that workflows verified through this method outperform unverified ones by an average of 11.94%. Additionally, the LeanEvolve component enhances workflow performance by 7.47%, establishing a new paradigm for formal verification in AI agent systems.
Verified workflows in Lean4Agent outperform unverified ones by nearly 12%, showcasing the power of formal methods in enhancing LLM agent reliability.
Equipping Large Language Models (LLMs) to execute reliable multi-step workflows has become a central challenge in artificial intelligence. Despite recent advances in LLMs' agentic capabilities, most agent systems still lack formal methods for specifying, verifying, and debugging their workflow and execution trajectories. This challenge mirrors a long-standing problem in mathematics, where the ambiguity of natural languages (NLs) motivates the development of formal languages (FLs). Inspired by this paradigm, we propose **Lean4Agent**, to the best of our knowledge, the first framework that uses Lean4, a dependent-type FL to model and verify agent behavior. **Lean4Agent** launches **FormalAgentLib**, an extensible Lean4 library for formally modeling and verifying agent workflows' semantic consistency under explicit assumptions, and enabling localization of execution-time failures revealed by trajectories. Building on **FormalAgentLib**, we further develop **LeanEvolve**, which applies results in **FormalAgentLib** to revise workflows to enhance its capability. Extensive experiments on a hard problem subset of SWE-Bench-Verified and a subset of ELAIP-Bench across 5 leading LLMs indicate that the verification-passing workflows outperform the failing ones by an average of **11.94%**, and **LeanEvolve** further improves SWE performance by **7.47%** on average. Furthermore, **Lean4Agent** establishes a foundation for a new field of using expressive dependent-type FL to formally model and verify agent behavior.
CMU ML