Stolen Fireworks AI Key Led to Data Breach and Tampering of Claude Code Extension

Published by staff writer

An access key stolen from Fireworks AI was used to compromise Anthropic’s official Claude Code extension on Cursor’s marketplace.

What happened

Cursor revealed that an attacker obtained an access key originally issued to Fireworks AI. This key was exposed due to a breach affecting Fireworks AI. Using the stolen key, the attacker was able to copy data from a read-only source and subsequently tampered with the Claude Code extension listed on Cursor’s extension marketplace, introducing malicious elements.

Why it matters

The incident underscores the risks of access key exposure and how compromised credentials can lead to the manipulation of trusted software extensions. Such breaches can undermine user trust and potentially expose users to malicious code through popular AI tools and extensions.

Who is affected

Users of the Claude Code extension on Cursor’s marketplace and potentially customers relying on Fireworks AI services may be impacted by the breach and subsequent tampering.

What to watch

Monitor updates from Cursor and Anthropic regarding patching the compromised extension and any further disclosures about the breach’s scope.

What you can do

Users should verify the authenticity of their Claude Code extension version and promptly update or reinstall it from official sources as recommended by Cursor or Anthropic.

Source: The Hacker News · Original report published Nov 15, 2026. Analysis is based on the linked report.