Well, they could make algorithms to automatically "own" some server just by supplying an IP, but it would fail very often and it would not be reliable at all.
For example the algo would do a port scan, identify running services, then try to gather as much other info as possible and use all that to find an appropriate exploit in the database and then run it and so on.....
But this would be quite stupid and unreliable and would fail very very often especially on more unique and secure targets.
This project basically aims to develop the ULTIMATE script kiddie tool
And we all know that even though skiddy tools can be powerful and destructive, they are only usable on limited and narrow range or situations and if a situation that was not thought about when the tool was created comes around the corner, then the skid is screwed.
This project might be finished, but it will be nothing else than the ultimate skidy tool that automates common cyber attack scenarios.
Thats how I see it anyway.