This allows me to ship a static html site (S3 website) where the javascript functionality is limited to basic things, like toggling dark mode.
To build an "API", I crawl the files at compile time and generate json files in an /api folder, allowing me to fetch blog post metadata for searches that are processed on the client.
As icing on the cake, once the user has settled in, I use a service worker to pre-download/cache the all the posts in the blog site and allow for offline viewing. I have some tricks to avoid the annoyances of stuck sw caches.
Leads to a great user experience and maintainers don't need to have a backend portal - just write markdown and commit it straight to GitHub. Easy enough to teach people to use Typora and SourceTree