This is mostly a question about the code for: https://github.com/kitasuke/SwiftUI-MVVM/blob/master/SwiftUI-MVVM/ViewModels/RepositoryListViewModel.swift
I see that you are organising bindings in Outputs and Inputs... in general you are adding a level of complexity that is very difficult to understand and read.

Take as example this step in the bindingInputs() method:

    let responseStream = responsePublisher
        .share()
        .subscribe(responseSubject)

and this one in the bindingOutputs() method:

    let repositoriesStream = responseSubject
        .map { $0.items }
        .assign(to: \.repositories, on: self)

Why don't you just merge the two writing just:

    let responseStream = responsePublisher  // Taken from bindingInputs()
        .map { $0.items }                                      // Taken from bindingOutputs()
        .assign(to: \.repositories, on: self)         // Taken from bindingOutputs()

Which is the benefit of having it done in two steps? I can't understand why putting a subject as connection between these two parts, while you could have subscribed with the .assign directly from the responsePublisher.

(I took hours to understand what you were doing here :P )

thank you for sharing your code! this is one of the few examples that shows a real implementation with Combine!

Firstly thank you for sharing your code, secondly i couldn't help but noticing that RepositoryListViewModel creates multiple instances at launch and a bunch of others on scrolling.
is that a bug in Xcode or something else i should be aware when using swiftUI in production apps.