Not the usual serverless migration

I build mostly one-off tools that take some input, crunch the data, and output something back.

Take Neutrality.wtf , for example. It accepts a URL given by the user, and then fetches the page server-side (to avoid CORS), parses the target, performs fairly heavy analysis and replacements on the structure, re-serializes and outputs the HTML for display. This operation is fairly heavy, and cannot be done on build.