I’m using the Free version of Cursor but already have a paid subscription with Claude. Can I somehow benefit from this in Cursor or will I have to also pay for Cursor ?
As someone who pays for all three (openai, claude, cursor)
The most benefit is to pay for Cursor - you WANT the Agentic YOLO…
im still on the fence to pay for anthropic, even though I want their artifacts, and the way it handles them. The walled garden, and that it cant @web is downright criminal…
THough I cant stand OAIs UX wrt copy_paste and the fact that it constantly doesnt want to give me full files, and it still feels like fisher_price rules forced upon users…
As these have zero lock_in, pay the $20 for cursor, and try it for a month…
I think ill keep the OAI sub as I like to get different perspectives on most things, though…
I use Claude a heck of a lot less now that I have Yolo, which is wonderful.
Thanks for your answer
I’m still pretty new to Cursor, so don’t know what you mean by “Agentic YOLO”, but I think I’ll try to start out with one month of Cursor and see where it brings me.
I was not sure if I could connect Cursor and Claude and bennefit from my paid Claude in Cursor
I wish that we could invoke a tertiary chat window that is a sidestream API connect via our personal key - so it would still be read_in on context (I have already successfuly had composer share the same vscdb from two different sessions and it was aware of the other composer DB table in the file…
But would like to have a window that uses your own token pool and key in same cursor session…
I have paid (Pro) subscriptions to Cursor, [the-editor-formally-known-as-Prince], ChatGPT, and Claude. I’ve dropped Perplexity as redundant, and “Prince” is on thin ice, but I’ll keep the others.
I’m at 2x Cursor Pro. I could end up at 4x Cursor Pro or so. I don’t care, this is for my math research. I am completely stunned by the extent to which Claude 3.5 Sonnet understands my iPad math drawings, and computational group theory, and strategies for combinatorial enumeration. There’s no point to my using a weaker AI.
To answer your question, AI is probabilistic. One really needs to watch the Christopher Nolan movie “Memento” to understand AI’s memory model. One gets better answers varying the prompts and contexts, when the answers really matter. For this, I do general brainstorming sessions in my direct Anthropic Claude Pro subscription, and use Claude through Cursor when I’m actually touching code. For me, this is an optimum usage strategy.
I’m new here, but I’m getting the impression that standalone Claude is smarter. This is probably an artifact of context management, but even so that speaks to the advantages of “rotating crops” and using Claude away from the codebase for general strategy.
With that observation, I’d take this approach much further. A standalone Claude 3.5 Sonnet session should be to generate a very detailed prompt as a sidebar artifact. Then bring it to Cursor, free or paid, to code. Using AI is a managerial problem. Not everyone appreciates managerial genius, but one view is that whenever workers screw up, it’s the manager’s fault. The prompt is the programming language, albeit one that needs assisted compilation. One should be able to get by with dumber agents in Cursor if the prompt is written well. If not, go back and improve the prompt. Test the prompt on multiple platforms, with multiple target languages.
I should note it’s worth paying $4 more for monthly rather than annual for the base Cursor Pro subscription, to be able to add and drop multiples on a monthly basis. There’s game theory here. My advice would be to add Pro multiples every time one runs out, and subtract multiples every time one wasted them unused for a month. And don’t fret over having wasted them; the AI help is valuable. Just also use an Anthropic subscription away from Cursor when that makes sense.
(Sorry about the length. AI loves detail. I’m forgetting how to write for humans.)