With that said, I still love Cursor and it remains my daily driver. How about you? What's your current tool of choice, and why?

I can't just send a whole folder or codebase to a model if I don't know how many tokens it has.

Tools like Repo Prompt, Cline and Roo Code are ahead of Cursor in this sense; serious software engineers are starting to take a closer look at these alternatives, given how much more control they give.

They need to rename this to "credits" ASAP.

2. We still don't have context size insights within Cursor. With token-based billing, this has gone from feature request to absolute requirement. As a hard-core MAX model user, I need full control over my context and my cost.

1. They still call your monthly credits "requests". Now that we have token-based billing with MAX, they can charge you 0.1 or 25.7 requests. But you've actually only made 1 request. Saw tons of people online being very confused about this, understandably so.

HOWEVER, it's a big mess for multiple reasons.

Yes we got OpenAI 4o and so on, but I have a huge notes folder and I want all of it to fit in the context window, and I want it to think through its answer first.

@SmokeAwayyy has a good deep-dive on the phenomenon.
x.com/SmokeAwayyy...

Context is king! We need tools like this to make it easy to cram more valuable context into our models.

See the full report here: cdn.openai.com/pdf/2221c87...

Haha I'm liking this test more than the ubiquitous 'bouncing ball inside of a rotating hexagon' thing I keep seeing everywhere.

The list goes on and on. Mantine has more components than any other library I've used, and I've never had to build anything from scratch.

It might be overkill for small projects, but it's the perfect choice for a UI-heavy web application!

Tons of useful hooks