Yeah, I'd imagine that trying to build something like that for SCI would be very difficult. The problem would be the object-oriented aspects of the language, which I don't think the existing Visual Programming examples (Scratch, Snap, etc) have addressed yet. And thinking about it, it would be difficult to represent concepts like that with blocks.
AGI is not object-oriented and reads more like a script. There are no user defined procedures or anything like that. It simply executes from the start of the script to the end. In some ways it is actually even simpler than what is possible in Scratch and Snap (where it is possible to raise events and have multiple event handlers, running in parallel). In the latest version of Scratch, and also in Snap, it is possible to "build your own blocks". So I think they're already beyond what AGI is capable of. And with AGI there is a one to one relationship between the original source and the compiled format, whereas SCI would in theory start off in some form of any language (even AGI source perhaps) and compile to the SCI byte code.
AGI commands logically fall in to groups as well, and therefore the Scratch and Snap way of grouping blocks under different sections would work well. AGI's concept of global variables and flags matches what Scratch has for variables (although Scratch also allows variables scoped specifically to a sprite, which is quite interesting, but once again shows that Scratch is more complex than what we'd need for AGI).