Cursor: Stolen Fireworks AI Key Behind Trojanized Claude Code Extension
Cursor has disclosed that the compromise of Anthropic's official Claude Code extension, distributed through Cursor's own extension marketplace, stemmed from the theft of an access key. That key was exposed as a result of a separate breach affecting Fireworks AI, a third-party AI infrastructure provider. According to Cursor, an attacker leveraged the stolen credential to interfere with the extension, resulting in a trojanized version being made available to users.
This incident illustrates a growing risk pattern in the software supply chain: compromise of a downstream vendor's infrastructure or credentials can cascade into trusted distribution channels such as IDE extension marketplaces, package registries, or plugin stores. Developer tooling ecosystems are attractive targets precisely because users implicitly trust official publishers, and a single stolen API key or access token can be enough to push malicious updates without triggering the scrutiny normally reserved for unknown or unsigned software.
Extension marketplaces, in particular, have become a recurring vector for supply-chain attacks because they blend the trust model of app stores with the lower moderation and faster update cycles typical of developer tooling. When an attacker gains access to a legitimate publisher's credentials, they can bypass many of the usual red flags that security teams and automated scanners look for, since the malicious artifact appears to come from a verified, known-good source.
For organizations relying on AI coding assistants, developer extensions, or any third-party integrations that depend on API keys issued by external providers, this case underscores the importance of tightly scoping and rotating access credentials, monitoring for anomalous key usage, and treating upstream vendor breaches as a direct risk to downstream software integrity. Security teams should also apply integrity verification (such as checksums or signing) to extension updates where possible, and maintain the ability to quickly identify and roll back compromised packages across developer environments.
Defenders should audit which third-party services hold API keys or tokens with publishing or update privileges for internal tooling, enforce short-lived credentials and least-privilege scopes wherever feasible, and ensure incident response plans account for the possibility that a breach at one vendor can silently propagate malicious code through trusted software supply chains.
Source: The Hacker News
Read the original report ↗This is an original summary written by CyberSkii for reference. All rights to the underlying reporting remain with the publisher linked above.