Cursor has published its full account of the October security incident that ended with malicious copies of Anthropic’s Claude Code extension being served from its marketplace. In an update on 13 November, the company says the chain started at Fireworks AI, one of its inference vendors: an access key Cursor had issued to Fireworks was exposed on Fireworks’ side, and an attacker used it to copy data from a read-only storage bucket the two companies used to share data.
The bucket held historical coding-agent traces, mostly from 2025, from sessions outside Cursor’s Privacy Mode that were eligible for model-training use. Cursor puts trace exposure at under 0.05% of users — and says a small number of those traces contained API credentials that had not been redacted, exposing credentials for under 0.01% of customers. Using credentials taken from that internal data, the attacker reached some of Cursor’s SaaS tools and, on 29 October, placed modified copies of Anthropic.claude-code versions 2.1.312 and 2.1.313 directly into the cache that serves Cursor’s VS Code extension marketplace.
The malicious builds were available for about 11 hours, from 2:27 a.m. to 1:35 p.m. Pacific Time on 29 October, and Cursor estimates that up to 5,000 users who requested the affected cached versions may have received them. Cursor says Anthropic did not publish the modified copies, the malware never appeared on the upstream Open VSX registry, and its production infrastructure, official updates, release signing keys and customer repositories were not affected. The last attacker activity it has seen was on 1 November, and it believes the incident is contained.
Cursor says it has revoked the Fireworks key, rotated the exposed credentials and then all internal credentials, moved all inference traffic off Fireworks, expanded secret scanning and network restrictions, rebuilt its extension marketplace, redeployed production infrastructure and engaged external forensic investigators. One point of attribution remains open: that the key was “exposed by Fireworks” is Cursor’s account — Fireworks’ own security update of 9 November does not mention Cursor, and Fireworks has disclosed two separate October intrusions without saying which one exposed the key. If you installed the Claude Code extension from Cursor’s marketplace on 29 October, check that you are on an official version and rotate any API keys that may have been in agent traces.