Should I Create a New Icon Library from Scratch or Fork an Existing One?

There’s also the Feather Icons library, which is a bit smaller and might suit your needs.

It really depends on what you want to achieve! If your goal is to learn, then building everything from scratch could be beneficial. But remember, this area is already quite well-explored, and there are many existing solutions. If there's something specific you think is missing, consider enhancing those libraries instead. By the way, I usually go with Iconify for my icons—it’s pretty handy!

If you feel lost considering forking big repos, maybe it's best to avoid that route. Large libraries like Lucide and Phosphor have complex build setups that could consume a lot of your time just to figure out. Instead, take inspiration from them—for instance, look at their size and weight attributes, naming conventions, and their approach to tree-shaking ESM exports. This way, you can develop your unique take without getting bogged down in their complexities.

Forking these types of design projects can be tricky. If you're looking to expand on what's already out there, it might be better to contribute to those existing libraries instead of creating your own. A new icon library should really have its own unique style or features. So, I'd suggest starting from scratch in terms of design, but you could definitely borrow some of their distribution practices to save some time.

I say go ahead and build your own from the ground up. It’ll be a rewarding experience! If Lucide inspires you, check out their design guidelines for contributions. Following those principles can help you achieve a similar style or even innovate your own.