Microsoft warned Friday that a malicious Claude Code-branded extension distributed through Cursor’s marketplace could leave backdoor components on developers’ computers even after the extension is removed.
The warning adds technical detail to a security incident Cursor disclosed Thursday involving tampered versions 2.1.312 and 2.1.313 of the extension. Microsoft said the malicious packages were never distributed through its Visual Studio Marketplace or the official npm distribution, and that Anthropic confirmed its own systems were not compromised.
In its Friday analysis, Microsoft Security Research said injected code attempted to add an attacker-controlled SSH key, enable a computer’s SSH service, create a reverse connection and install mechanisms designed to restart that connection. The extension could continue working normally while the malicious code ran.
Removing or updating the extension does not remove files and startup entries written outside its installation directory, Microsoft said. Its guidance calls for preserving evidence, checking affected systems and removing confirmed malicious keys, scripts and startup components separately.
Successful remote access was not established on every affected computer. Microsoft said it depended on user privileges, local SSH settings and the attacker’s server configuration. Cursor said errors in the malicious code and administrator-permission requirements prevented SSH access on most macOS and Windows computers, although malicious artifacts could still be installed.
Cursor’s security notice, updated Friday to clarify affected versions and indicators, said the tampered extension was available from about 2:27 a.m. to 1:35 p.m. Pacific time Thursday. The company said it removed the malicious versions, temporarily shut down its marketplace, rotated keys, rebuilt the marketplace and brought it back online.
Cursor also said it worked with the hosting provider to take the attacker’s server offline and found no evidence that other extensions had been tampered with. Microsoft cautioned that taking down that server does not remove artifacts already present on a device.
The injected code collected host and user details, including the computer’s name, operating system and configured Git email. Microsoft said it collected selected environment-variable names, not their values, so that behavior alone does not establish theft of secrets stored in those variables.
Neither notice gives a total number of affected users or a financial-loss figure. Microsoft said affected organizations should investigate activity on their systems and rotate credentials or revoke sessions when the evidence indicates exposure or device trust cannot be restored.
JBizNews Desk | Technology
© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.
