There are CPU test ROMs available that draw information to the LCD and output to the serial port. Using the serial port it would be possible to feed the information given by those ROMs into a simple script that then verifies whether the tests succeeded or not upon commit using Github Actions with a Docker container running a MTA server (would this be a first instance of a MTA resource testing itself on CI?).

There might also be tests for other aspects of the Game Boy but that should be further researched. If so, this issue should be expanded to include them.