Hackers Compromising Developers with Malicious VS Code, Cursor AI Extensions

The report documents research revealing that attackers can publish malicious VS Code and AI-IDE extensions which execute on launch, detect security tools, exfiltrate environment variables containing credentials, and deploy a Merlin command-and-control agent; it also shows sandbox/geo-fencing evasion of Microsoft’s vetting and minimal verification on OpenVSX, creating a significant developer-targeted software supply chain risk.