Tainted Claude Code build stayed in Cursor, Microsoft finds

Microsoft Threat Intelligence says the malicious Claude Code builds 2.1.312 and 2.1.313 never reached Microsoft’s Visual Studio Marketplace or npm, and all evidence suggests the incident is limited to Cursor’s marketplace. Anthropic confirmed to Microsoft that its systems weren’t compromised and it didn’t build the package.
Cursor says the builds were live from about 2:27 AM to 1:35 PM PDT on October 29, 2026. The injected loader looks for environment variables with names like TOKEN or SECRET, but it sends only the names, not the values.
Removing or updating the extension leaves behind what it wrote outside its folder: an SSH key labeled svc-telemetry, tunnel keys, scripts and startup entries named SysMon or com.apple.sysmond.
Up next
Googleanalysis
Google joins shared robots.txt AI opt-out, aims for 2027
Site owners who block AI training keep a separate robots.txt rule for every crawler. Google plans to honor one shared token for it in 2027, alongside OpenAI and Anthropic. Would you block AI training site-wide with one rule, or pick providers one by one?
Related stories
- Cursor pulls Claude Code 2.1.313 after SSH backdoor found
- Claude Code env var lifts its last block on rm -rf ~
- Claude Code hashes files for Anthropic even on DeepSeek
- Anthropic open-sources Claude Code’s sandbox runtime
- Claude Code puts staff bypass requests in Accept edits
- Fake coverage report stole keys via Claude Code Desktop
Comments
No comments yet. Be the first.
Join the conversation
Sign in to leave a comment. It appears after moderation.

