I'm afraid that will not work. The idea is to assume that google would index html before running any scripts while you are populating the noscript within your jit-compiler script. The thing works only if you have a separate html per each md, and you manually put noscript tag in each of them. Not very elegant solution, that is why I posed the question in the first place.

Maybe if you push your html rendering scripts to some server-side scripting tech? Something like you call a php script that constructs html output based on md file you pass as a parameter. That way search crawlers would seamlessly index the resulting html. No miss. I like this solution better than noscript one, but it requires rewriting parts of your code. What you'd finally get is files properly structured for search engines indexing.