Undoing changes and file state

[Charlie_Hess]

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Probably the biggest issue with cursor is its memory of file state. There are so many times I click undo on a limited changeset in a file I’ve already clicked “Keep All” on several times before, and cursor blows the whole file away. The number of times I have to type to my agent “cursor deleted the file, pls recreate it” is ridiculous. Any sustained session inevitably falls out of sync with file state and before you know it, very, very strange things start happening. There are also new bugs with the IDE strangely deciding to reopen every plan it has ever worked on? Idk guys, stability on this thing is dropping the last few versions. If I were you I’d drop everything and do a dedicated sprint on file state. That has to be bulletproof for an IDE and it is decidedly not.

Steps to Reproduce

Yeah of course I have no steps to repro. All I can say is if you’ve used the app enough you’re definitely hitting this bug.

Operating System

MacOS

Version Information

Version: 2.4.31 (Universal)
VSCode Version: 1.105.1
Commit: 3578107fdf149b00059ddad37048220e41681000
Date: 2026-02-08T07:42:24.999Z
Build Type: Stable
Release Track: Default
Electron: 39.2.7
Chromium: 142.0.7444.235
Node.js: 22.21.1
V8: 14.2.231.21-electron.0
OS: Darwin arm64 25.2.0

For AI issues: which model did you use?

Opus 4.6

Does this stop you from using Cursor

No - Cursor works, but with this issue

[deanrie]

Hey, thanks for the report. A few things:

Undo after “Keep All” deletes a file: We know there are issues in this area. There was a recent regression where undo didn’t roll back internal state correctly, which caused a mismatch between what the agent “thinks” is on disk and what’s actually there. Some fixes are already shipped. Can you check that you’re on the latest version? (Help > About Cursor, or check for updates)

If it’s still happening on the latest build, it’d really help if next time you grab the Request ID: click the three dots at the top of the chat > Copy Request ID. That lets the team trace exactly what went wrong in that specific session.

Plans unexpectedly reopen: This is a known issue and it’s already been passed to the team. No ETA yet, but reports like this help raise its priority.

For now, Restore Checkpoint in the chat timeline can be a safer way to roll back when undo is acting weird. It restores files from a known snapshot instead of trying to undo step by step.

Let me know if things are better on the latest version, or share the Request ID next time it happens.