I was tinkering with an idea this past weekend after developing a slightly complex multi-step workflow, would love some feedback on it. I know it's a long read, the basic idea is:
Is it smart to build a workflow that knows very little and relies on an API to directly serve Alfred results and decide what paths to take? The workflow ultimately serves as a basic input collector, and cumulatively passes the information back to the API. I think it's definitely interesting, but am interested to hear what people think.
https://blog.joe.codes/api-controlled-alfred-workflows-an-experiment/