Cursor Traces Breach to Fireworks Key Leak; Fake Claude Code Extension Reached Up to 5,000 Users
Cursor published on Nov 13 the findings of its October security incident: an attacker used an access key issued to inference vendor Fireworks to copy coding-agent traces from a shared storage bucket, found unredacted credentials in them, and on October 29 placed a tampered Claude Code extension in its VS Code extension marketplace cache.

What happened
The incident started at Fireworks. Cursor says an attacker obtained an access key it had issued to Fireworks, and that the key was exposed by Fireworks; the attacker used it to copy a read-only storage bucket the two companies shared, which held a limited set of coding-agent traces, mostly from 2025 sessions outside Privacy Mode, a small number of which contained credentials that had not been redacted.
The attacker used those credentials to reach some of Cursor’s SaaS tools and obtained archived non-production source repositories, customer support data and email conversations, product usage metrics, summaries of some coding-agent conversations, and credentials users had sent to the support team. On October 29, the attacker used a stolen credential to place tampered copies of Anthropic.claude-code versions 2.1.312 and 2.1.313 directly into the cache that serves the extension marketplace.
The tampered builds sat in the cache for about 11 hours (2:27 a.m. to 1:35 p.m. Pacific Time on October 29), and at most 5,000 users who requested the affected cached versions may have received them. Cursor says Anthropic did not publish these copies, the issue is not in Anthropic’s product, and the marketplace source Open VSX never contained the malware; Microsoft confirmed the malicious packages were not distributed through its Visual Studio Marketplace or npm. Cursor says the last attacker activity it has seen was on November 1, and it was blocked.
Key facts
- Origin: An access key issued to Fireworks was exposed, and coding-agent traces in a shared read-only storage bucket were copied
- Credentials: Fewer than 0.01% of customers had API credentials exposed; fewer than 0.05% of users had traces exposed
- Tampered builds: Anthropic.claude-code versions 2.1.312 and 2.1.313 sat in Cursor’s extension marketplace cache for about 11 hours on October 29
- Exposure: At most 5,000 users may have received the tampered builds; Cursor says follow the steps in its October 30 security notice, as uninstalling alone may not be enough
- Not affected: Prompts and code in Privacy Mode, customer code repositories, production systems, the Cursor app and its official updates, and release signing keys
- Response: All inference traffic moved off Fireworks, the key deactivated, all internal credentials rotated, the extension marketplace rebuilt, external forensic investigators engaged
- Attribution: “Exposed by Fireworks” is Cursor’s account; Fireworks’ November 9 update does not mention Cursor
Why it matters
The incident links a vendor key leak, credentials inside agent traces, and extension-marketplace tampering into one chain, making traces a new carrier of credentials; Fireworks’ account has not yet been matched against Cursor’s version, leaving one part of the attribution open.
CompaniesAnthropic
Grab todayOpenAI Codex for Students: $100 in Free CreditsHIMIAO Picks →Follow HIMIAO Horizon
No sign-up · Unsubscribe any timeNotifications
A ping on your phone or computer when the brief is out. No email needed.
Comments
Related
- Nov 10Anthropic Reports About 100 Korean Enterprise Accounts Leaked, Regulator Confirms
- Nov 9Anthropic Launches Claude Enterprise Security Center; Inference Hooks Go GA
- Nov 8Claude Code Settings Leak Put Real Secrets in 1,122 Public Repos
- Nov 1Trojanized Claude Code Extension Hit Cursor's Marketplace, Affecting Up to 5,000 Developers
- Oct 13Anthropic Says Anti-Distillation Safeguards Cut Extraction by 90% or More
Today
Nov 17 Tuesday- Brief
- GitHubGitHub Weekly Vol.6: 7 Projects Worth a Look This Week
Tested or read — weeklyAll →
- Quiz
- Call
Latest news
All →- Nov 17US Commerce Department Closes "Cloud Loophole" for Advanced AI Chips
- Nov 17OpenAI Launches Science Blog as GPT-6 Astra Leads Proof of Candidate Lattice Standard Model Definition
- Nov 17Apple Smart Glasses Said to Debut at WWDC 2027 With No Display, Leaning on Siri and iPhone
- Nov 17AI Labs Race to Hire Counterintelligence Staff: Anthropic 33 This Year, OpenAI 22
- Nov 17Trump Backs Huang's Return to Nvidia: 'He Learned a Big Lesson'
- Nov 16Kuo: OpenAI and Anthropic to Buy About 26M Google TPUs in 2028–29
™