Short term opinion about Cursor

With the native Claude chat (in the browser), I’m able to get the needed results even with free Claude tier limitations.

Perhaps to adjust to Cursor (pro tier), I need to use special prompt techniques or do something else. Currently, I notice the context drifting significantly. Maybe Cursor sends too much information, or I need to organize my requests in the AI chat section differently.

I’m not sure. I was trying to use Cursor to work more efficiently…

I’m not even comparing the JetBrains (paid) workflow with Cursor… It seems to me that IntelliJ IDEA + Claude brings better results.

If you folks feel the same, please share your tips or guide on how to switch from IDEA + Claude to Cursor.

Thanks!

My favourite current features of Cursor is the ability to have it index docs so that you can @ them and it can find relevant pieces for answering questions - just RAG built in really. I think Cursor tab completion (copilot) is really good - but I haven’t used another in a while so I’m not sure if it’s better than others. At the end of the day, if Cursor can understand the code base really well and not make mistakes when writing code, it will be all you need right? Not there yet but getting closer.

1 Like

Thanks for replying!

Did you compared your own results from Claude browser against Cursor results?
With the same query Claude in browser able to bring better results than Cursor. Maybe Cursor send to much context( bigger than 5 chunks available in Claude browser).

When I’m attaching related files by myself in Claude browser - I’m able to get result what I need. With Cursor @File feature - no.

Reading the feedback section of the forum I see that people also find the Cursor Claude responses became worse neither usual Claude in the browser.

Oh sorry I think you’ve misunderstood- I meant reference docs by ‘docs’ e.g. tensorflow API docs

Just so that I can look through them easily via cursor.

I haven’t compared Claude in Cursor to Clause in the browser