CMake has a GUI which is nice, but it's likely because the CMake settings is highly formalised.<p>But Autotools does have a lot of commonality so if we restrict the autotool configuration into a structure (and give enough par-sable annotations) then it might just be possible to do something similar.
Why not use a LLM bridge here? Either as directly interpretation -> execution<p>Or as a tool to find a way of to formalize, to find distinctive areas where it's possible and where not. Just a thought...