Cursor Breach Exposes Developers to Malicious Claude Code Extension
Cursor confirms it suffered a security breach that let hackers distribute a malicious Claude Code extension through its marketplace after stealing credentials from AI infrastructure provider Fireworks AI. The attack exposes up to 5,000 developers to malware that installs persistent SSH backdoors. Cursor is an AI-powered code editor built on Microsoft's open-source Visual Studio Code that lets developers write and debug software using AI assistants.
According to Cursor's incident report, attackers obtain an access key stored at Fireworks AI and use it to copy data from a read-only storage bucket shared between the two companies. The bucket contains historical AI coding-agent traces from users who have not enabled Privacy Mode, some with unredacted API credentials. Using these stolen credentials, the hackers access additional Cursor systems and steal data from several SaaS tools, including archived source code repositories, customer support conversations, usage metrics, and credentials customers previously shared with support. Cursor states fewer than 0.05% of users have coding-agent traces exposed and fewer than 0.01% of customers have API credentials compromised.
Cursor's monitoring systems detect the unauthorized activity, prompting the company to revoke the Fireworks access key, rotate exposed credentials, and investigate further access. As the investigation progresses, Cursor temporarily shuts down some internal and production systems. The incident highlights risks when sensitive AI training data and credentials are shared with third-party providers without adequate redaction.