DeepYardDeepYard
V

VeriSkill

Self-evolving framework enabling LLM agents to learn and refine program verification skills

Open SourceFree

About

VeriSkill is a research framework that enables LLM agents to autonomously improve their program verification capabilities through self-evolution. The system allows agents to distill verification skills from execution trajectories and feedback, creating a continuous learning loop for automated formal verification tasks. Designed for researchers exploring autonomous skill acquisition in code verification domains.

Details

Language
Patterns

Tags

autonomouscoding-agentframeworkopen-sourceself-hostedevaluation