Thank you. I saved the prompt and documentation in Google Keep. 

Should we consider designing a robust and strict way to interact with
currently running instances with a debug port, for any external commands to be
implemented in any external node, in any language -- provided they just
understand the structure of the program state and operations? 

Also, we could look at documenting and making a "development kit" for any
external actors, that could read the program's state, and introduce events,
and add external commands --- in such a way, that the new external commands
COULD also be transmitted via any kind of means to the external service? 

We'll in other words need to be able to register anonymous functions/closures,
OR, perhaps better, register commands, (and routines?), that are in strictly
defined and controlled way, delegated by the VM, to perhaps one of these: 

1. Mapping a certain command to a REST call

2. Mapping a certain command to an UDP datagram packet to host:ip

3. Leave open the possibility to configure other custom backend-services
later, when the engine gets further developed

There would naturally need to be timeouts/safe error handling when the REPL
runs these commands

What do you think?