Now I am starting to think, that C3 will benefit a lot by allowing developers to access everything in the engine easily and make changes, regarding the way you wanna takle the project (lower/higher level). In some way you can do it now too, but it's more the hacky way. Sure SDK is for that, but it's only one part of it. I recently tried a JS-plugin - it was very easy to include an external js-library and use it, basically the way you bind js-code with construct events through this plugin. Open API's could also be used easily.
In this context Construct could be seen not only as an html5 game engine, but more like an intuitive IDE for web-apps-development and more. Would be cool if it would have a built in functionality to render/edit DOM elements and draw with canvas. Standard construct render (layout/scene) could be included into the certain DOM elements.
So basic workflow for more lower level development would be:
1) create bindings between you JS-HTML-CSS in Construct3 (functions, objects, arrays, variables, callbacks, divs, html-objects, css-objects-values, etc)
2) use events to create higher level logic to bring all js/html stuff to life
This will also allow to use databases and other backend services more easily, without writing a plugin or behavior.