What kind of protocol do we need for remote hacking? - evaluating code and receiving the results - handling remote errors with restarts etc - redirecting output streams This would be enough for e.g. live hacking a server application.