Cursor affected by the Fireworks AI intrusion incident; a malicious version of the Claude Code extension was once listed on its marketplace
Cursor受Fireworks AI事件波及,惡意版Anthropic Claude擴充程式於Cursor市集上架
Cursor announced on November 13 that, due to fallout from the Fireworks AI intrusion incident, a malicious version of the Anthropic Claude Code extension was listed on its VS Code extension marketplace on October 29, and an estimated up to 5,000 users who requested a specific cached version may have received the malicious version. The investigation showed that attackers obtained access keys issued by Cursor from Fireworks AI's systems, and that data copied from a shared read-only storage bucket contained a small amount of unmasked authentication information; fewer than 0.01% of customers had API authentication information leaked, and through this the attackers entered some of Cursor's SaaS tools, obtaining some archived source code repositories, customer support data, and summaries of some coding agent AI conversations. Cursor removed the version about 11 hours after it was listed, suspended the extension marketplace service, and published an advisory and a check script containing indicators of compromise; no attacker activity has been observed since November 1, and it considers the incident contained.
The Fireworks AI breach cascaded to Cursor, and a malicious version of the Claude Code extension was briefly listed on its marketplace, which can be used to examine the transmission path of AI supply chain risk.