I have been thinking for a long time about learning and applying reinforcement learning to see how many levels can player go with or without mouse click bullet assist. Game code is simple, small enough to understand and make modifications.
A sample game based on modified getting started tutorial. It implements minimap , unity cinecamera like movement, different hud & game layers, custom collision box, button events, mouse click bullets with physics etc.
This can be played here: https://quickgrid.itch.io/ninja-sample
Movement done with arrow keys. O and P keys are used for some fight animations which are not implemented and finally mouse click will shoot cats.
