Howdy folks,
Has anyone here thought of reading memory to enhance the bot capabilities? I've started messing around with reverse engineering the data structures in-memory, and pretty quickly came up with this:

It wouldn't take too much effort to add troop building, raiding, etc. It also has the advantage that you can see the exact composition and layout of a targeted base, including the resource in each collector, storage, etc as well as clan castle troops. I've also started getting some basic information on traps, which would be impossible from a pixel-based bot as they're typically invisible.
If anyone has started heading down the same road and is interested in exchanging information, please feel free to PM me.
