The Daily Byte
Breaking
Welcome to The Daily Byte — real stories from across Pakistan.
Saturday, 21 November 2026 · Pakistan

The Daily Byte

Stories from Pakistan that deserve to be heard

Cursor Confirms Fireworks Key Breach Behind Fake Extension

By The Daily Byte ·15 Nov 2026 ·4 min read
A laptop showing lines of code in a dark room

Cursor has confirmed that a security breach at its artificial intelligence vendor Fireworks AI led to a booby-trapped copy of Anthropic's Claude Code extension being distributed through its own extension marketplace, in an incident report published on Friday 13 November.

The AI code editor company said the attack began when hackers obtained an access key it had issued to Fireworks AI, a provider it used for model training and inference. Cursor said Fireworks exposed the key, and the attackers used it to copy data from a read-only storage bucket the two companies shared.

That bucket held historical coding-agent traces, mostly from 2025, collected from users who had not switched on Privacy Mode, the setting that keeps prompts and code out of model training. Some of those traces contained unredacted API credentials, which the attackers then used to reach further into Cursor's systems.

What the attackers reached

Cursor said fewer than 0.05 per cent of users had coding-agent traces exposed, and fewer than 0.01 per cent of customers had API credentials compromised. Using the exposed credentials, the hackers stole data from several of the company's software-as-a-service tools, including archived, non-production source code repositories, limited customer support information and email conversations, product usage metrics and customer contact details, summaries of some AI coding-agent conversations, and credentials customers had previously shared with the support team.

Cursor said its monitoring systems detected the unauthorised activity, and it revoked the Fireworks access key and rotated exposed credentials. But the attackers had already obtained a credential that let them distribute extensions through Cursor's marketplace.

The fake extension

On 29 October, the hackers used that credential to publish malicious copies of Anthropic's Claude Code extension on the cache that serves Cursor's VS Code extension marketplace. The tampered builds, versions 2.1.312 and 2.1.313, were available for about 11 hours before Cursor removed them.

Cursor estimates that up to 5,000 users who requested the affected extension versions during that window may have received the malicious builds. The modified extension ran malicious JavaScript that gathered system information, installed an attacker-controlled SSH key, tried to switch on SSH access and set up persistence, in an attempt to install a persistent backdoor on developers' computers.

Security researchers found coding errors in the malware that likely stopped it working on many Windows and macOS systems where SSH was not already enabled. Cursor stressed that Anthropic's legitimate extension was not compromised, and neither was Open VSX, the extension marketplace it draws from. The fake copies were placed directly on the cache serving Cursor's own marketplace.

What users should do

Cursor said the last attacker activity it observed was on 1 November, and it believes the incident is contained. The company said it has moved all inference traffic away from Fireworks, revoked and rotated exposed credentials, rebuilt its extension marketplace and redeployed its production infrastructure as a precaution.

Customers who installed or updated to the affected Claude Code extension during the attack should follow Cursor's remediation guidance. The company warns bluntly that uninstalling the extension alone may not be enough, and affected users should update to version 2.1.314 and check their machines for the attacker's SSH key. Cursor said its production infrastructure, official application releases, customer code repositories, and the code and prompts of users with Privacy Mode switched on were not affected.

The incident is part of a wider run of October attacks on AI developer platforms that cybersecurity company Wiz has called "Black October", in which attackers targeted development infrastructure to steal credentials giving access to AI models. The same Fireworks AI breach also exposed data at Uber and Notion. Cursor's full account is in its incident report as covered by BleepingComputer, and The Hacker News has a detailed account of the chain of events.