Show HN: Cloakroom – Keeps secrets away from agents (github.com)

Cloakroom has been launched as an open-source tool designed to enhance security for AI coding agents by masking API keys, passwords, and personal data on a user's machine before these agents can access them. This tool is compatible with several coding agents including Claude Code, Codex, Gemini CLI, and Cursor, and it utilizes a local detection model to swap sensitive information with placeholders during operations, ensuring that real values are securely managed in an encrypted vault. The significance of Cloakroom lies in its ability to prevent accidental data leaks and protect sensitive information during AI development—an increasing concern as AI tools get integrated into workflows. Notably, Cloakroom demonstrated its effectiveness in testing, successfully preventing leaks of personal details and keys while allowing only a minimal number of false alarms. With features like health checks, alerts for false masking, and a simple installation process across multiple operating systems, Cloakroom stands as a valuable addition to the toolkit of AI/ML practitioners looking to secure their development environments against potential threats.