So is the idea that you create commands that call different HTTP backends? Do the backends need to support the user-input-as-header stuff, or is this translated

So is the idea that you create commands that call different HTTP backends? Do the backends need to support the user-input-as-header stuff, or is this translated to call arbitrary HTTP backends? Please give some more info about the use-cases you are trying to support. I'm interested, but not quite interested enough to read the code to find out if I want to use this.