'CursorJack' Attack Path Exposes Code Execution Risk in AI Development Environment (infosecurity-magazine.com)

• Malicious MCP deeplinks as entry vector
• Code execution requires user approval
• Exposes trust gap in AI dev tools

"A newly identified attack vector called 'CursorJack' exploits malicious MCP deeplinks in the Cursor IDE to trigger user-approved code execution. The attack highlights a critical vulnerability in the security models of AI-assisted development environments, where seemingly innocuous links can lead to severe security breaches. This incident underscores the need for stricter security protocols and user awareness, particularly in AI development contexts."