This toolkit helps you build agents of any windows game you have using python. It uses native windows APIs to capture a game window at very high frame-rate. Then uses block level tesseract to provide context awareness for building state machine. This helps build high-quality deep reinforcement learning bot without any knowledge of game internals (e.g. No C++ Hooking/Assembly Patching or DLL injection).

Aim of this project is to provide an end-to-end toolkit which can be used to build bot for any game. All that will be required, is to build a state machine adapter which helps navigate through screens and lets agent(s) play the game and feedback a reward mechanism.

Note: This project is a Work in Progress!