Executive Summary
A stolen Fireworks AI access key allegedly enabled the compromise of Anthropic's official Claude Code extension on Cursor's marketplace. Organizations using AI coding agents, extensions, or third-party AI services should review their software supply-chain and credential controls.
What Happened
The compromise of Anthropic's official Claude Code extension on Cursor's extension marketplace was the result of the theft of an access key that was exposed in the wake of a breach targeting Fireworks AI. "An attacker obtained an access key we had issued to Fireworks," Cursor said in a Friday update. "The key was exposed by Fireworks, and the attacker used it to copy data from a read-only
Affected Products
- Claude Code extension
- Cursor extension marketplace
- Fireworks AI
Vulnerability / Attack Technique
Stolen access key and AI supply-chain compromise
Technical Mechanism
According to the summary, an attacker obtained an access key issued to Fireworks AI after the key was exposed by Fireworks. The attacker used the key to copy data from a read-only source, and the incident resulted in the compromise of the Claude Code extension on Cursor's extension marketplace.
AI Security Category
LLM03:2025 Supply Chain
Part of the AI Supply Chain and AI SBOM topic.
Relevant Frameworks
RealGround Analysis
AI agents and LLM applications often depend on extensions, marketplaces, model providers, and external service credentials, so a compromise in one provider can propagate to downstream users. Teams should maintain an AI-specific software bill of materials, verify extension provenance, scope third-party credentials narrowly, and monitor read activity and package changes.
Recommended Controls
- Maintain an inventory of AI extensions, marketplace packages, model providers, and third-party integrations.
- Use short-lived, narrowly scoped access keys with read-only permissions where possible, and rotate or revoke exposed credentials immediately.
- Require signed packages, verified publisher identity, integrity checks, and approval workflows before deploying AI extensions.
- Monitor extension updates, credential use, and unusual data-copy or read activity across AI development environments.
- Isolate AI coding agents and extensions with least-privilege permissions, restricted network access, and sandboxed execution.
Related RealGround Service
AI Agent SBOM and LLM Mapping · or start with the free AI Agent Security Scanner.
Sources
thehackernews.com: https://thehackernews.com/2026/11/stolen-fireworks-ai-key-led-to-cursor.html
Summary and analysis written by RealGround from the source above; last updated 2026-11-15.
