Cursor pulls malicious Claude Code extension
Cursor removed a tampered Claude Code 2.1.313 extension after it attempted to install an SSH backdoor and exfiltrate developer-environment data. Independent analysis identified the malicious artifact in Open VSX, while its status in other distributions remained unverified.
This is a serious supply-chain warning for AI coding workflows: editor extensions run with enough local access to turn a trusted marketplace into a direct path to developer infrastructure.
- –The payload collected host details, added an attacker-controlled SSH key, and attempted reverse tunneling.
- –Cursor said the malicious package was available for roughly eleven hours and may have affected up to 5,000 users.
- –The incident highlights why extension provenance and marketplace isolation matter as AI coding agents gain filesystem and shell access.
- –Developers should verify installed versions, inspect persistence mechanisms, rotate credentials, and treat unverified extension updates as executable code.
DISCOVERED
1h ago
2026-10-30
PUBLISHED
1h ago
2026-10-30
RELEVANCE
AUTHOR
SocketSecurity