I've been using Cursor as my primary editor for a month. Before this, I was a VS Code loyalist with Copilot, a pretty good setup that I thought would be hard to beat. Cursor beat it on day one. But it's not without problems. Here's my honest review.
What Cursor actually is
Cursor is a fork of VS Code with AI deeply integrated into the editing experience. Your extensions, keybindings, and settings transfer over. It looks and feels like VS Code, so the switch is painless. The difference is in how AI is baked into every interaction, not as an add-on sidebar, but as part of the editing flow itself.
The killer feature: Cmd+K inline editing
Select some code, hit Cmd+K, type what you want changed, and it edits inline with a diff view. This sounds simple but it's transformative. Instead of copying code to ChatGPT, reading the response, figuring out what changed, and manually applying it, you describe the change and see it applied directly in your file.
I use this constantly for small refactors. "Extract this into a separate function." "Add error handling." "Convert this to async/await." Each of these takes 5 seconds instead of 30. Over a day, that adds up to easily an hour saved.
The chat panel knows your codebase
Cursor's chat panel can reference your entire project. You can @-mention files, and it indexes your codebase for context. This means when you ask "how does the authentication flow work in this project," it can actually answer based on your code, not generic patterns from training data.
This is miles ahead of Copilot Chat, which has limited context awareness (I wrote about that experience in my first month with Copilot). I've had conversations with Cursor about multi-file refactors where it understood the relationships between components, services, and routes without me having to paste everything manually.
Tab completion is scary good
Cursor's autocomplete predicts not just the current line but multi-line blocks based on context. It regularly suggests the exact function I was about to write. The prediction quality feels a step above Copilot, though this is subjective and varies by language. For TypeScript and Python, Cursor's completions are consistently better. For Go, it's closer to a tie.
The rough edges
Performance is the biggest issue. Cursor is noticeably slower than VS Code, especially on larger projects. I work on a monorepo with about 200K lines of code, and I get occasional lag when navigating files or running searches. The AI features add overhead, and it shows.
The AI sometimes gets confused across files. I've had instances where an inline edit in one file made assumptions based on outdated context from another file. It's not frequent, but when it happens, the suggested edit looks correct at first glance and introduces a subtle bug. You still need to review everything carefully.
Pricing is aggressive. $20/month for Pro, which is what you need for the good models (GPT-4, Claude). The free tier uses smaller models and the quality difference is significant enough that the free tier feels like a demo rather than a usable product.
Some VS Code extensions behave oddly. Most work fine, but I've had issues with a couple of debugging extensions that don't play well with Cursor's fork. Nothing deal-breaking, but worth testing your critical extensions before committing.
My workflow changes
Before Cursor, my AI workflow was: code in VS Code, switch to browser for Claude/ChatGPT when I need help, paste context, read response, apply changes manually. Now it's: code in Cursor, use inline editing for small changes, use the chat panel for larger discussions, everything stays in the editor.
I still use Claude directly for architecture decisions and complex debugging. Cursor is excellent for the moment-to-moment coding experience, but for big-picture thinking, I want a dedicated conversation with a model I trust for reasoning.
The verdict
Cursor is the best AI coding experience available right now. The inline editing alone justifies the switch from VS Code + Copilot. If you want to see how it stacks up against the competition, I did a detailed Windsurf vs Cursor comparison. If you're already paying for Copilot ($10/month) and ChatGPT Plus ($20/month), switching to Cursor Pro ($20/month) might actually save you money while giving you a more integrated experience.
I'm staying with Cursor. The performance issues are real, but the productivity gains outweigh them. This is what AI-assisted development should feel like, not a chatbot in a sidebar, but intelligence woven into the editing experience itself.