VSE and motiontracking workflow help

Hello,

I recently started to use Blender for editing videos. I usually have just 2 or 3 clips which I cut directly into several parts and rearrange them in the VSE. Effects like crossfades and tuning with the adjustment layer work fine.
Now I want to mix in more features of blender, like adding rendered strips and the camera tracking. So my question is, “What is a good workflow to use these features?” Have you any advices or tutorials at hand? Not specifically how to use blender features like for example the video proxy.
Some questions I have are:

  • Is it better to cut each clip into single files and load them into the VSE?
  • Which format should I use for this cut-down clips? (in most cases, I am not using audio)
  • Do I use a new scene for each clip?
  • If I use camera tracking, do I render a new clip and use it in the VSE or can I do it in one step?
  • If I want to use the compositor for finishing the final video, do I need to render it first and then reload it (as images?) – what about sounds?

I am eager to learn how to set up blender and use a good workflow for polished videos.

I have searched a bit for this script but with no hint where to get it. Do you have a name or location where to download / read a documentation?