Debug mode is actually amazing 🔥

By EagerOwl7486 · DISCUSSION · About Cursor

This mode is amazing at adding its own instrumentation then testing and confirming its hypothesis. I love it! Good job Cursor 👌

139 upvotes · 7 comments

Comments (7)

ShadowSpark326: Yes very good for debugging . Probably the best front end feature cursor has introduced to date. Love the hypothesis and testing logic ingrained into it. Very robust

PixelSurfer27: I've tried it several times and I think I'm to stupid to use it

HappyFox759: I haven’t used it yet. What’s the difference from just telling cursor something doesn’t work? It worked fine so far.

NeonGamer522: Works much better when you include the expected vs actual behavior upfront — the model forms a testable hypothesis instead of randomly probing. Something like 'clicking X should do Y but instead does Z' gives it a concrete target.

BoldVibe231: I really like it as well. I mostly use Claude Code nowadays, but I missed this feature when I first wanted to debug something with it, so with Cursor's help of its own tool, I built a CC skill that works mostly the same way as Cursor's debug mode: [https://github.com/kerdofficial/claude-debug-mode](

SwiftEcho377: someone How does it compare from a token usage standpoint to the code review feature in an active agent? I'm getting stuck in issue loops for large feature rebuilds (we're using cursor on existing codebase vs. a built from scratch with cursor code base). I'll use plan, then use agent to build and th

StormPulse179: Can it also debug dotnet apps ?

More discussions about Cursor on HonestUse