Your exactly right, this is the way I am suggesting, but the key with what I am trying to create is to make it even simpler than HAML, remember an old one that vanished a few years ago... KISS.
Nevertheless, this is the right way to go, but the syntax to HAML is still a little more cryptic than what I am suggesting. There are less things to remember when keeping the building process reduced to one line entries.
Also the interactive help can work easier that way by staying context sensitive when based on the mode of the module being developed. And because it is not simply a application compiling the content but an entire LFS operating system based around networking in general, it can be as simple as it can get.
In other words you do not need to complete scripts entirely before you can see the results. Why not create a table along with graphics etc., and have a peek. Why not build sections of Perl that you can test and store wherever building the scripts like Lego.
So thanks for your positive reply, I do get them a lot. What is important though is you do understand what I am suggesting. I did start compiling a LFS a few years back but got bored - this was about 12 years back. That was because I had no need for it, now I think it is ideal.