It is just extremely concerning that your development team is shipping new features and at the same time is failing to address a pretty serious issue (although not affecting most of the people) which has been known for more than a month now.
You don’t find that a decision to set Cursor to scan my entire filesystem is a bit controversial? Okay, I understand that this could be just a mistake in the filters passed to ripgrep, but then why not roll it back immediately? This is what I would have done in my team, if this code for some reason had even passed a code review.
One of your engineers stated here that the problem will be fixed in Cursor v2.4 release. I’m on version 2.4.21 and the problem is still there.
Workarounds? Well, I have tried limiting access to the shares, but Cursor still finds its way there. I’ve tried using AppArmor to restrict access, but no luck. Currently what works is to rename or move the rg binary before I start Cursor, so it doesn’t run a full scan, however that bricks the File Search and the Agentic mode, even after I restore the binary, so not really a usable workaround. I’m thinking that creating a sandbox for Cursor might actually work, but I don’t have time to tincker with this.
This is extremely frustrating.