My experience is that once you understand the (high level) data structure you're manipulating by using git, you have already made 80% of the way there. I barely know basics of git's interface and have to use tools like `git gui` for anything more complex in my everyday workflow, yet I can manage just fine even when presented with complicated problems, since once I know from the data structure perspective what I want to do, finding a way to do that in git's interface by searching manpages and/or the Web is hardly ever a big issue.