Using AI to help configuring RouterOS and scripting

Can you provide some more info on what you trained The Dude chatbot on support page? Was it just help docs, or other things?

My question here is what use case do folk expect or want from a LLM connected to RouterOS?

For example, is it configuration generation, help to understand a particular syntax/configuration, orchestrating configuration, log analysis, etc... Ideally where someone things chat interface be better than using traditional tools.

Reason I ask is I'd like to expand my VSCode LSP+notebook to provide CoPilot more context internally.

There are a few different ways to help CoPilot perform better within my TikBook VSCode extension... but IDK what things actually be useful. But I do know CoPilot perform much better when providing more contextual data to CoPilot – like the export configuration and various "print"/status things from a connected RouterOS & interactively providing the help doc also help.

So kinda asking for input on what tasks folks might even want to use an LLM with as providing too much context data to CoPilot/LLM uses more tokens (and likely reduces accuracy if given context windows) & also guide agent prompting to CoPilot from the extension.