@wdormann @briankrebs @GossiTheDog It was blown to giant proportions to start with. For some reason some people that pays for the Cursor LLM harness want to use the Claude Code LLM harness inside the other LLM harness (??) from this proprietary open source mess that Musk thought was worth $60 billion. Cursor has some kind stupid proxy in front of Open VSX that they use to look like VS Code to defend their value, and this was abused to publish a fake release to a couple of thousand users. This release contains a backdoor that seems to be broken. It's literally the least important thing people need to worry about right now, so I understand that Anthropic, that literally has nothing to do with it, didn't feel any big urge to comment when Cursor already had taken 100% ownership of the compromise and issue.
ffforward@infosec.exchange
Indlæg
-
Anthropic got compromised. -
Anthropic got compromised.@mttaggart @GossiTheDog
There never was an official 2.1.313 version. That release was entirely fake. So to avoid confusion (which also gives Cursor users something to upgrade to) the next official release after 312 was instead 314. -
Anthropic got compromised.@GossiTheDog @mttaggart I'm not sure, there was no reason for this to explode at all, Cursor took full ownership pretty fast, and did not claim upstream was responsible in any way. If the first posts about the issue wouldn't incorrectly have blamed upstream, they wouldn't have needed to comment at all. There can't be that many users of Claude Code on Cursor, why would you want to run the extension for one paid harness when you likely already have an subscription via Cursor, specifically to use their harness?
Also Anthropic didn't publish a "fix", to me it just looks like they skipped a version number because they knew that releasing under that version would create even bigger confusion. -
Anthropic got compromised.@GossiTheDog I don't find any indications that it was anything but a compromise of Cursor systems (proprietary plugin ingestion it seems) , I don't find anything that the official extension or OpenVSX would have carried the malicious version. What evidence have you found that it's bigger?
In addition VS Code has an default extension update block for 12h, so it wouldn't have been affected.