Cursor Security Incident: Malicious Claude Code Extension Distributed After Fireworks AI Key Compromise
An exposed Fireworks AI access key led to a supply chain attack on Cursor, resulting in credential exposure and the distribution of a malicious Claude Code extension.

By Elena Rostova, Cybersecurity
3 min read
Cursor, the AI-first code editor, has disclosed a significant security incident stemming from a compromised access key issued to its AI infrastructure partner, Fireworks. The incident led to the exposure of sensitive user data and the distribution of a malicious version of the Claude Code extension through Cursor's VS Code extension marketplace.
The attack originated when an attacker gained access to an API key that Cursor had provided to Fireworks. According to Cursor's official blog post on November 13, 2026, the key was exposed by Fireworks, allowing the attacker to copy data from a read-only storage bucket used for sharing information between the two companies. This bucket contained limited coding agent traces, primarily from 2025, from sessions where Privacy Mode was not enabled. These traces were intended for model training purposes.
Crucially, a small fraction of these traces, less than 0.01% of customers, contained unredacted API credentials. The attacker leveraged these compromised credentials to gain unauthorized access to several of Cursor's internal SaaS tools. This breach allowed them to access archived non-production Cursor source code repositories, some of which contained business contact information. Additionally, limited customer support data, email conversations, product usage metrics, contact details, summaries of some coding agent conversations, and credentials sent to the Cursor support team were also compromised.
Further escalating the incident, on October 29, the attacker used a stolen credential from Cursor's internal systems to upload malicious, unauthorized copies of Anthropic's Claude Code extension directly to the cache serving the Cursor VS Code extension marketplace. Cursor explicitly stated that Anthropic was not involved in publishing this compromised extension.
Upon detecting the unauthorized activity, Cursor initiated an immediate response. They deactivated the compromised Fireworks key, launched an incident response investigation, and began a comprehensive process of revoking and rotating credentials. The company also temporarily disabled various internal and product services as a precautionary measure, restoring them only after ensuring their security. Cursor has confirmed that it has directly contacted all affected customers with specific details regarding the impact on their data.
This incident highlights the critical importance of supply chain security in the AI development ecosystem, where the compromise of a third-party partner's credentials can have far-reaching consequences for user data and platform integrity.
Sources
- 1Cursor Security Incident Updatecursor.com