As much as this approach might work well for side-projects, it would definitively lead to a few pain points for bigger teams or projects. The main being that you are binding the display of your data (frontend) to your server (backend). In many cases this means multiple deploys for a single purpose, and also multiple people involved to solve a problem that might actually be very simple.

I've also seen many times this leading to the anti-pattern of building an endpoint to every single new page/feature. Once you reach hundreds of endpoints where some of them are very similar, things starts to get messy.

I believe in most cases the API should empower the user to express and manipulate the data in every way the domain allows, even if it means building complex urls/queries sometimes.