Cisco donates Project CodeGuard for AI coding agents

The author introduces Project CodeGuard, an open-source, model-agnostic framework of security rules and skills donated by Cisco to the Coalition for Secure AI. The framework translates security guidance into IDE-specific rules and works across different moments in the software development lifecycle to prevent AI coding assistants from repeating known vulnerabilities. Additionally, the author demonstrates a detection-to-prevention flywheel that automatically synthesizes new prevention rules from security gaps found during codebase scans. Open article →