CNnotator: LLM-Guided Memory Safety Annotation Synthesis

CNnotator is a hybrid testing and verification tool that utilizes large language models (LLMs) to automatically synthesize memory safety annotations for legacy C code. The OpenAI o3 reasoning model achieved a 90% success rate on initial attempts and 97% overall in generating CN specifications for small-to-medium C programs, while GPT-4o achieved a 65% success rate on first attempts. This advancement indicates the potential for AI-assisted annotation to enhance memory safety in existing C codebases, facilitating migration to safer programming languages.