Malicious Claude Code extension on Cursor planted SSH backdoors

A malicious version of Anthropic's popular "Claude Code" assistant for VS Code was recently distributed through the marketplace of Cursor, an AI-powered code editor. According to reports from BleepingComputer, the compromised extension was downloaded by up to 5,000 developers before being removed, exposing them to malware that installs persistent SSH backdoors. The malicious software was available for approximately 11 hours.

Incident Details

The attack targeted users of Cursor, an AI-native code editor, by publishing a fraudulent version of a legitimate VS Code extension for Anthropic's Claude. The malware embedded within the extension was engineered to create a persistent SSH backdoor on the developer's machine, which would grant attackers ongoing, unauthorized access. The incident highlights the risks associated with third-party software marketplaces, where malicious packages can slip through vetting processes. The extension remained active on the marketplace for about 11 hours, during which time it was installed by as many as 5,000 users.

The Evolving Threat Landscape

This supply chain attack occurs in a context of increasingly powerful AI tools being used for both development and malicious purposes. For example, recent research has shown that open-weight models can autonomously build full software exploit chains with no human intervention in under 48 hours. As AI becomes more integrated into developer workflows and consumer products, the potential attack surface for supply chain vulnerabilities expands. Developers must remain vigilant about the sources and integrity of the tools they integrate into their environments.

Recommendations for Developers

Developers using extensions from any marketplace, especially those outside of official, heavily moderated repositories, should exercise caution. It is crucial to verify the publisher and review permissions before installing new tools. In this case, users of the official VS Code marketplace were not affected. Teams should implement security policies that govern the use of third-party extensions and dependencies to mitigate the risk of supply chain attacks. Regularly auditing installed tools and monitoring for suspicious network activity can also help detect and respond to potential compromises.

Additional sources: - the-decoder.com - techmeme.com