Hey Cursor team, which AI models are best suited for Debug mode?
Hey! 
I’ve put Debug Mode through its paces with a ton of different bugs recently. In my experience, GPT-5.x is the clear winner. It’s honestly really good at forming a hypothesis and then actually validating it compared to the others.
1 Like
I used debug mode yesterday to find many sneaky regressions causing unit tests to fail.
It’s incredibly helpful.
question: what’s the best way to tell cursor to write logs to a location that is a writeable within my docker container?
Hey @srhyne
Haven’t tried this myself, but it looks like you can configure ndjson.relativeLogFile in your VSCode JSON settings to change where the logs end up. Give it a try and let us know if it works?
1 Like
Nice! Thanks!
Thanks for sharing that, I’ll try it out!
From what I can tell, Debug mode actually adds tracing statements to your code, that call back into a locally hosted service of some kind. I believe this is how the mode tracks what’s going on in your code.