A malicious extension that carried the Claude Code name and ran inside the Cursor code editor reportedly planted SSH backdoors, giving attackers a potential route into affected systems.
Supply chain attack · Developer tools
A fake Claude Code extension in Cursor reportedly planted SSH backdoors
A malicious extension that carried the Claude Code name and ran inside the Cursor editor was reported to install SSH backdoors, potentially giving attackers a route back into affected systems without the owner's knowledge.
How the reported attack works
A simplified flow based on the reported mechanism
Step 1
Disguised extension
Uses the Claude Code name to look legitimate
→
Step 2
Runs in Cursor
Executes inside the AI-focused code editor
→
Step 3
SSH backdoor
Installs remote access the owner may not see
→
Step 4
Potential access
Attacker can regain entry to the machine
What the reporting establishes
The extension used the Claude Code name to appear legitimate.
It ran inside the Cursor code editor.
It was used to install SSH backdoors.
SSH backdoors can let attackers regain access unnoticed.
What remains unclear
How many users were affected.
How the extension was distributed.
How long it stayed available.
The exact infection mechanism.
Any response from Anthropic, Cursor or the marketplace.
Developer checklist
If you installed an extension claiming to be Claude Code or any AI coding add-on, check it now.
Verify the publisher
Confirm the listing comes from the official developer.
Review permissions
Look closely at what the extension is allowed to do.
Check SSH keys
Look for unexpected authorized keys on your machines.
Audit remote access
Review remote access configurations for changes you did not make.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...