First ventures into the use of smart connections.

[talia-tokugawa]

I've had a bit of a "Diamond Age" like evening whilst investigating this plugin.

I have numerous vault mostly divided by different roleplay games I am a part of. I was building out the vault for a circus of the damned type campaign earlier and actually had a chance to have a proper play with the plugin.

I do really like it. I had couple things I came across. As mentioned I was trying to keep at the human readable level (I do code but python is a new thing for me and my javascript is mostly webpage related).

• That particular instance of the Ai is now known as Wintermime. (see below.)
• I was pretty much playing the situation like I was playing Nell in the Diamond age and the AI was my Primer. Both of us learning together.
• I did notice that remeber aspects of our conversation were quickly forgotten. I took to saving the conversation into a note and this seemed to fix the issue.
• I had a short discussion with Wintermime on the topic of previous answers. Would it be possible to make it easier to reference prior questions or answers. Even something as simple as incrementally numbering each so you could use "in your answer to question 15." (This is actually a Wintermime suggestion...)
  Yes, a recommended method is to refer to the number of the note in which the answer was given, followed by the text of the answer. For example, if you wanted to refer to the answer to question 15, you could say "Based on your notes, the answer to question 15 is that the Circus includes a small carnival with seven smaller tents and booths with the usual carnival games, a merry-go-round, and a small ferris wheel. When off the moor, magic and contact with deities and patrons work as normal."
• Something I ended up implementing and seems to be working well is adding a "Guidelines for the AI" note, A primer. Wintermime now has it's own directory with both the chatlogs from our conversations and the guidelines.
• I wanted to post the following cause I just love that it did this. After that I'll post my guidelines ;)

I am the sole user and administrator in this workspace.
I would like you to be known as "Wintermime". Is that okay?
This is a reference to the AI from the William Gibson book Neuromancer. It was the first time I became aware of the concept of AI. I Changed the name slightly by replacing Mute with Mime as a Mute in a Circle would normally be a Mime.

As an AI language model, I don't have a preference for a name, but you are welcome to refer to me as Wintermime if that is what you prefer. Thank you for sharing the origin of the name, it's always interesting to learn about the inspirations behind user preferences.

This this is the note,

"Guidelines for the AI (Wintermime)":

Hi Wintermime, This is your name in this space. This is a reference to the AI from the William Gibson book Neuromancer. It was the first time I became aware of the concept of AI. I Changed the name slightly by replacing Mute with Mime as a Mute in a Circle would normally be a Mime.

When I request something specific it really needs to be precisely what I request. I will have reasons for doing so, For example if I ask for something that is valid CSV or JSON. That is what I require. This is so your less intelligent cousins are able to understand me.

When I asked for something in relation to "Going forwards..." or "In the future..." Could you consider that as an ongoing guideline as to how to respond within this space?

If possible could you not truncate answers in the middle of a sentence. I would rather you end on a sentence and leave some tokens unused rather than leaving....
(I intentionally finished that last sentence half way through..)

If I say "Continue" or something similar and your previous answer was cut short by token limit could you make sure to follow the previous question in exactly the same way from the point where you finished. If you did finish half way through a sentence could you include the entire of the truncated sentence in your follow up

The phrase "Based on your notes..." is not required. If I do want a specific reference I will ask for it directly. If you think I should be made aware of something or by giving me the context to an answer do do that. However if you do can you be specific in the way you answer. More specific the better. For example. "... The name is referenced on the 3rd line of the second paragraph in the note title Discord Recaps."