Cursor Crashes Constantly After Latest Update (code: ‘5’) — Cannot Work at All

Support Bug Reports
27

This exact error started for me yesterday morning. I’ve tried the recommended fixes, not working. For me, it was the latest software update that did it. Is there a way to undo that update? This is so frustrating.

UPDATE - I ran the following and it’s working now! Yes, I got this from ChatGPT but it’s working so I don’t care.

Proper fix (full reset of corrupted Cursor databases)

  1. Quit Cursor completely

Make sure it’s not running in the background.

Cmd + Q

Then confirm:

ps aux | grep cursor
  1. Delete the corrupted IndexedDB + LevelDB stores

Run this:

rm -rf ~/Library/Application\ Support/Cursor/IndexedDB
rm -rf ~/Library/Application\ Support/Cursor/Local\ Storage
rm -rf ~/Library/Application\ Support/Cursor/Session\ Storage
rm -rf ~/Library/Application\ Support/Cursor/GPUCache

Then also remove the provider index:

rm -rf ~/Library/Application\ Support/Cursor/User/index
  1. Restart Cursor

Cursor will rebuild all the databases on launch.

This is safe — it only deletes cache and indexing data.

If it STILL fails (rare but happens)

Your Electron profile itself is corrupt. In that case do a clean rebuild:

mv ~/Library/Application\ Support/Cursor ~/Desktop/Cursor_backup

Then launch Cursor again.

That forces a fresh profile, which fixes almost every crash-from-power-loss case.

Why this happens

Cursor stores a ton of AI state locally:

  • codebase embeddings

  • chat history

  • agent memory

  • workspace graphs

All of that lives in LevelDB.

When a write gets interrupted, the database becomes unreadable and the provider hangs.