I've been working on a pretty complex JavaScript project and I'm running into a problem. The AI coding assistant I'm using, Cursor, keeps running out of context when trying to process the entire project, which leads to it losing track of earlier code and making less helpful suggestions. I'm really on the lookout for AI tools or strategies that can handle larger context windows without losing track as the project scales. Ideally, I'm hoping to find something that can manage large codebases or entire projects without hitting memory or token limits, understands multi-file structures and can handle references across files, works well with JavaScript (like Node and frontend frameworks), and offers a good interface for iterative development and debugging. Has anyone tried any tools or workflows that stay coherent during long-term or multi-file project editing?
3 Answers
Feels like a lot of these AI tools just end up being fancy garbage for the price. Not sure how useful they really are for larger projects.
I think the best tool for managing large projects is your own brain! It's amazing how a human can navigate complex code better than many AI tools right now.
I heard about something called 'Dropstone' that claims to have unlimited tokens. Has anyone actually tried it? Is there a downside?
From what I've gathered, Dropstone can run open-source models which is different from being locked into usage-based pricing like Cursor. If you have the hardware, it could cost around $12 to run it yourself and potentially avoid those token limit issues. Just curious if anyone else has experience using it long-term!
Absolutely! This is solid advice for any coding issue - sometimes we just need to trust our instincts.