Rapid: An LLM-Driven Framework for Autonomous Vulnerability Remediation Across Heterogeneous Operating Systems Environments

Timely and efficient vulnerability management remains a critical challenge in cybersecurity, particularly as organizations struggle with heterogeneous operating system (OS) environments where remediation steps vary significantly across different platforms. After vulnerabilities are identified, organizations face substantial delays in developing and implementing effective patching solutions, often due to the gap between generic vulnerability descriptions and OSspecific remediation requirements. The Large Language Models (LLMs) based framework presented in this research generates customized OS-specific remediation instructions by synthesizing data from the National Vulnerability Database (NVD), vendor-specific documentation and web scraping while incorporating a verification system. This framework leverages LLM models for instruction refinement, focusing on technical accuracy, and potential system impact assessment across different operating systems. Initial testing across diverse enterprise environments demonstrates an 84% accuracy in the generated instructions. This approach not only accelerates patch deployment and reduces operational risks but also provides organizations with a scalable solution for maintaining security postures in heterogeneous computing environments.