Nodes have an input (function argument) and an output (return value - possibly another function, which would form a node with two inputs).
Function application is just connecting the output of one node to the input of another. You have builtin operator nodes for set projection (which also covers attribute selection), set union and other standard functions.
From there it's just building new nodes from other nodes and in the end you get a neat visualisation of, say, what your system configuration looks like without having to dig through the file system and read the code spread throughout various files.
You can see the resulting output for any input easily by creating and plugging in dummy values, or perhaps the inputs that were used to build the current system.
Throw in Ctrl+F, inline documentation and online searching...
Sunbeam City is a anticapitalist, antifascist solarpunk instance that is run collectively.