To launch the application, do
docker run -ti jouster/bug_ai
This will put you inside the container in order to use the bugAI. What is happening is that based upon the times of when you give it input, the bug will take an average of how long it took for all instances and use that as how long it will take to kill itself off (it expects that based on the current time to get food, it will kill itself off knowing that it should instantly duplicate itself at that point in time). Each time you feed the bug, it doubles. Each time the death timer is reached, the bug count is halved. The program will end when the bug count reaches 0 or until a large enough number of bugs becomes present.
The scripts utilize parallel processing in order to successfully pull this stunt off as the death timer and the replication script need to be able to communicate to each other simultaneously. Each sub-script makes modifications to two additional files, one that contains the average, the other that contains the number of bugs.
If you want to see the script itself, simply use
docker cp *BoxID*:/ /path/to/place