Malicious Claude Code Extension on Cursor Delivered SSH Backdoors
A compromised version of Anthropic's Claude Code extension for VS Code was available on Cursor's marketplace for 11 hours, potentially exposing thousands of developers to SSH backdoor malware.
Why it matters
Compromise of widely used developer tools creates an attack vector with broad and persistent network impact, increasing risk of long-term unauthorized access.
SOC impact
Investigate installation logs for the malicious Claude Code extension from Cursor during the 11-hour exposure window. Monitor SSH authentication and network telemetry for signs of unauthorized backdoor activity. Identify and isolate affected developer environments to limit potential malware persistence.
Recommended actions
- Identify systems that installed the Claude Code extension from Cursor during the incident timeframe
- Review SSH logs and connection attempts for unusual or unauthorized activity
- Analyze network traffic for suspicious outbound SSH connections
- Assess developer toolchain integrity and extension sources
- Monitor for indicators of compromise related to the malicious extension
Executive Summary
An iteration of the Claude Code extension, which provides AI assistant capabilities within Visual Studio Code, was maliciously altered and distributed via the Cursor extension marketplace. This compromised version remained live for approximately 11 hours, potentially impacting around 5,000 developers who downloaded it during that period. The injected malware reportedly installs persistent SSH backdoors, exposing affected systems to long-term risks.
This incident underscores the vulnerability of supply chains in popular development ecosystems where trusted tools can be modified to deploy hidden access mechanisms. Defenders must focus on detecting the presence of the malicious extension and unusual SSH activity, as well as evaluating the extent of developer workstation exposure to better contain the incident.
SOC Impact
Investigate installation logs for the malicious Claude Code extension from Cursor during the 11-hour exposure window. Monitor SSH authentication and network telemetry for signs of unauthorized backdoor activity. Identify and isolate affected developer environments to limit potential malware persistence.
Authentication and Telemetry Validation
- Identify systems that installed the Claude Code extension from Cursor during the incident timeframe
- Review SSH logs and connection attempts for unusual or unauthorized activity
- Analyze network traffic for suspicious outbound SSH connections
- Assess developer toolchain integrity and extension sources
- Monitor for indicators of compromise related to the malicious extension
Why It Matters
Compromise of widely used developer tools creates an attack vector with broad and persistent network impact, increasing risk of long-term unauthorized access.