Cursor or Claude Code?

I recently stopped paying for cursor and moved fully to Claude Code. It is a step up in pricing but the output is usually more accurate and requires fewer iterations vs what I'd have to do in Cursor with the same prompt. I also have some really interesting data from our own user base on Cursor code quality with Claude vs Claude Code.

• Claude Code on average produces less overall code reworks by close to 30% and gets things right in the first or second iteration. Whereas Cursor with Claude 3.5 and 3.7 still tends to produce higher code churn.

• We've also seen a lot of our own customers move to Claude Code and Codex fully while sticking to using VSCode.

• Claude code also tends to produce code which follows existing patterns and has higher modularity by default. We know this because on our reviewer gives less recommendations on modularity and code abstractions for customers of ours that use Claude Code and Codex.
  
  I still use cursor as my go to code editor for making non-complex side-projects but for internal use we our engineers are still very much divided, half on Cursor and half of us on Claude code.
  
  My current stack is:
  IDE:
  @VS Code with Claude Code
  MCP:

  @GitHub and Context7
  Reviewer and Codebase maintenance:
  @Optibot

Nah bro. Cline + gemini cli hock. 2x Free. Best models. No context cap. 60 calls per min. Brrrrrr. Breakdown: https://www.producthunt.com/products/google?comment=4693656

I found a free stack that is currently working pretty well:

1. Visual Studio Code

2. @Cline (100% free).

3. @Google AI (free tier).

I have been using Cursor and I like the flexibility to switch between coding by myself and asking AI to do it, which Claude Code CLI might not offer

Depends on your coding style I guess. What style do you prefer?

Claude Code is remarkably better at returning good code. It's so much better that I find myself wondering how Cursor and Windsurf are not able to be as good given the underlying model is the same (maybe?).

The downsides to Claude code are:

• It costs way more and you have very little control over how long it takes to do things and how much it thinks

• I don't like doing everything via command line. Having the diff view in the IDE with Cursor/Windsurf is much nicer for me. Claude Code has plugins for the JetBrains IDEs but they are still clunky compared to the VS code options.

Some things that make Claude Code amazing are:

• Integration with GitHub, ex tag it on issues, pull requests etc and it will work async

• MCP support for all kinds of goodies. Some fun ones are Sentry and Linear, ex go fix this bug or go build this feature

I’ve used both before quite extensively. Have since opted for the Claude Max plan because Claude just seems to work with my workflow and the results are mostly pretty good. A few things I’ve come to love:

• Even though it’s a CLI tool, it’s pretty integrated into my IDE. When installing the CLI, it automatically installs the Jetbrains or VS Code extension, so you do get to see and manage the changes side by side in your IDE.

• The MCP functionality is pretty basic but it just works. With a good system prompt built over time, it knows when to run puppeteer to visually test my changes. It knows when to create and apply prisma migrations. It even knows when to access the Internet for more up to date documentation or ways of implementation.

• The output is really good. It’s built a few stable iterations to features that have been better than the previous versions than I have built manually in the past (I am an engineer)

• The context window management isn’t amazing, but it works. When you hit the context window limit it summarises what it’s been working on and discards the previous context window, allowing you to continue indefinitely

• You can manage how it processes information. It has a few keywords that you can use (like “ultrathink through your solution”) to use the thinking model more or less effectively, which helps you balance usage

• The limits are reasonable. For the $100 per month, I use Claude pretty much all the time (often as an unbound agent) and I hardly hit the limits. They have the idea of a session (a five hour window) where you can send something like 250 messages.

• It’s really all about how you use the tool. I’ve set up a pretty in-depth system prompt (a local CLAUDE.md file) that describes design, testing, and coding patterns, rules to follow, etc.) it also manages its own plan.md file as it builds so I can tweak a plan on the fly.

I think it’s worth it. As a dev turned people manager, I love that I can dive into code if I choose, but I haven’t had to write lines of code in months. Even considering the $200 20x Claude max plan (the current max plan has opus model usage run out in about an hour, and it is a better model)

I prefer having a user interface over using the terminal, because it allows me to better understand what the AI has changed and gives me a greater sense of control.

Gabe, totally with you Cursor has changed the way I approach small builds and quick experiments. Being able to work directly in a GitHub repo without breaking flow is a huge win. It’s less about flashy features and more about frictionless momentum.

Haven’t personally used Claude yet, so can’t compare, but if Cursor is helping you learn and ship, that’s already a big win in itself. Would love to hear how it stacks up if you end up testing both!