Hi there,
I’m trying to write e2e tests across several services in different folders within same repo, and I am considering using Cursor to improve my productivity.
I was wondering what is the extent to which Cursor can make modifications. Can it propose whole, large-scale changes across several microservices, or is it more scoped to copilot-esque help?
Also, is it able to reflect and converse about architectural decision/pros and cons before committing to a direction?
Thanks a lot for your answer!
Cheers,
Philippe
Both of what you are asking about, to a degree, is possible with Cursor!
Regarding modifications, the feature you’ll likely want is called “Composer”. Using composer, you can select a series of files or folders as context, and propose a change you want made within those files or folders. The AI will construct a plan of action and execute it, giving you the option to accept or reject changes once it’s done. It also has the ability to make new files when needed to finish its prompt. However, to avoid the AI making changes in irrelevant files, you do have to specify the context it can work with - it currently doesn’t work on your whole codebase without you pointing it towards the right files.
Composer is exactly for your described purpose - features and changes that need aligning across multiple files or, in your case, microservices.
Regarding the reflection, you can use the AI chat feature to discuss architecture, similar to ChatGPT, but with the added context of your codebase. That can work by automatically looking at your codebase as context, so may be better to figure out how to implement something before you do it. You can then pass the decided plan of action into Composer for it to execute.
The great thing about Cursor is there is a solid free tier available, so give it a go! You can import your extensions from VSCode and you should feel right at home.
A great place to learn more about the features is on the website to learn about them, and the docs site to master them! If you’ve have any issues or suggestions, feel free to drop them back on this forum!
Website: Features | Cursor - The AI-first Code Editor
Docs: Cursor - Build Software Faster
Thanks for your thorough response @danperks!
You convinced me to give it a shot! 
I’ll be happy to see how Cursor can supercharge my work! 
Sounds good. As I say, there’s a solid free trial available so see how you get on!
If you like it, the normal plan is only $20 a month and you can cancel anytime (but you do get a discount if you really like Cursor and buy an annual subscription).
Also, you can put in your own API keys for OpenAI, Gemini and Anthropic to be charged at-cost each time you use the AI, but the $20/month usually works out cheaper for daily users and lets you use all the models from any platform.
Feel free to raise a new post if you have any questions or issues and we’ll help where we can 
1 Like