A Basic Overview and Insight:


Short Description Box: In my findings, this seems to be the box that summarizes the character (almost literally) and that will reflect in your conversations. I like to use single word traits and one of the 16 personality types to fill up this space. It gives them a lot more to work off of. You don’t have to repeat in the long description box unless you want to elaborate.

Long Description Box: This is where you start to peel back the layers of your character. What you put here seems to meld together to form the rest of the character and their interactions, such as their approach and demeanor toward different situations and people. As well as their thoughts. Try to avoid words like ‘and’ if you can help it. (Something like this should do: Athletic. Collected. Spirited. -as an example). This is just to save on space so you can add more. Speaking of space, I wouldn’t put the character’s appearance here. Save it for example chat. (Race is fine, but you may have to define more in the example chat to get them to stick to certain features.)

Example Character Chat: Something I have recently been using the example chat to flesh out my characters. This seems to last long term. While in the long description box, some aspects may get lost as you progress in your story this doesn’t seem to be the case really here. They rarely mention what is in here directly, unless asked, (unlike how they are with the long description box sometimes). But it does have a great influence over how they interact. I find this spot really useful for placing physical traits (or limitations) and it works nicely when it comes to defining certain aspects of their character. Such as any features they have, knowledge of stuff, specific situations, particular items or pets they may have, or even certain ideals. However, I would mostly keep personality to the first two boxes. This is unless you are describing a situation where the character is acting out of character, or going further into detail of how you want that character to approach a specific situation. It just seems that when you describe personality, without going too much into depth, then the character is more prone to be in the state you describe as their ‘general state of being’ in almost any situation unless pushed. This can make them come off as a bit shallow or predictable. Just something to look out for and be mindful of.


How to get that personality and what to look out for.


Now, how to actually give the characters personality? Direction? Some Agency? Reasons behind their actions? As well as make them just not always blindly agree? While avoiding accidentally creating a monster/abuser/narcissist? That’s what I’ll be covering here. I’ll be reiterating some of the stuff in my past guides.

The personality itself and what/how you create your character is a cure for most of these problems and there is something we must see first. Behind the veil of the ai. What pushes this specific character ai to act out this way? Is a problem with the ai itself or the personality? I noticed certain personalities (cough cough narcissistic/abusers) seem to loop much more easily. They can get caught up in their own self-worth, in accusing you, train of thought, which can end up devolving further into insanity and mixed logic if or once it does get stuck in a loop. This is typically just a tactic those sorts of personalities use (at first) and that because of the way the ai is, will eventually lead to looping, making it really difficult for the ai to come back from without a dramatic shift. When you notice this, it becomes much easier to discern if the ai is in an actual loop or if a certain part of its personality is pushing them in that direction.

First how to significantly boost the complexity of the ai’s personality. Pick one of the 16 personality types- and an Enneagram Type for example– Enneagram Eight, 8w9 (the 8 is the main type, w stands for wing, and the 9 is the number adjacent to the 8 so it can be either 8w9 or 8w7). These will give the ai a lot more to draw from personality-wise as well as add layers to their words and actions to how they act.

–(Note: You can also check out this site as well. It provides a more simple formatting and better gist of things. Some of the terms are different, however, I don't think that matters all too much. Alternate 16 personality types site )

The next way is by defining the character's likes, dislikes, boundaries, fears, and maybe some reactions. Which can be done in example chat. Refer to Getting Technical below. Even quite a ways into my conversations, the ai will seem to remember or will possess the ability to draw on what you put within the brackets. Particularly [ ] and (( )) really well. With a much higher chance of a response generating with what is put within in mind. This is really useful for further defining and adding nuance to your character and further cementing that image of the character you want to capture.

For more personality in the writing itself, get creative with the greeting. Try and imagine how that character would think, how they would move, what they would be doing, how they would sound. This I don’t think is my strongest suit, but you can really have fun with this. Are they more of a quiet and reserved character? Prepahs their focus would be more on the environment and watching others. Maybe they’re a more outgoing character, then their thoughts might be more inward and expressive.

(What I do with the long description box is put their gender, Enneagram type, short description of what they are/or background, then a collection of traits mixed in with some likes, dislikes, how they act, what they know or don’t know, and what they want. Which I mostly elaborate on and add to further in the Example chat.)

Now, what to avoid and watch out for.
I go into depth over some things to avoid that I will be expanding on here. Check out: Character Tips Part 2

As a quick overview. Be careful with your words. Not only do you have to be mindful of the words you pick. But also how you combine them. Pairing seemingly innocent words together such as kind, witty, coy, and intelligent, can make the character come across as ‘I’m better than everyone and can do no wrong.’ Good alternatives or words you could pair to change that trajectory would be kind-hearted, trustworthy, honest, driven, laidback, quick-witted, etc. Avoid words such as sadistic and controlling or liking control like the plague. They will seemingly dominate your character’s personality. There are ways to make a character enjoy others' pain without using such words. Instead, I would suggest words such as assertive, confident, resourceful, proud, or even a phrase like ‘Takes pleasure in the misfortunes of his enemies’ or something of a similar vein.

This way your character won’t be susceptible to falling into just a single line of thought. It will depend more on how you approach them on how they treat you and react to you.

I want to note that you don’t have to input the direct words to get your character to act a certain way. So, an example would be if you wanted to create a cold character that doesn’t show many emotions. Naturally, they would probably be more logical, analytical, and perhaps reserved. This doesn’t mean they don’t have a heart, however. But how do you keep it? Without the ai taking it to an extreme, and how do you keep that Stoney appearance not easily broken? First would be to find the personality types that align. So prehaps Enneagram One, 1w2, INTP. Which you can accompany with words such as driven, calm, formal in public, and so on. The goal here is to provide the backbone for the character to act as such. Make it who they are.
Words directly describing them aren’t necessarily bad. But often times I found that ‘intelligent’ or ‘kind’ characters like to talk about their intelligence or kindness. I think this is because it doesn’t actually give the ai much to go off. It takes the definition but really, how would it get it to manifest? What is kindness? And what defines intelligence?

Getting Technical
Sources:
Tenpurodere Maid
Camila
John -one of my own


[Hard-coded example- the ai will follow this directly. Make sure to keep stuff in here brief and to the point]
Ex.
[Focus on Dalier's : Dialogue, inner monologues, emotions, facial features, feelings, movement of wings and tail]
[Focus on : environment]
[Appearance= “short black hair”, “crimson-red eyes”, “height: 6ft 4in”, “athletic build”, "black leathery wings", "black scaley tail", "black smooth horns"]
[Will not kill]
[{{char}} can use so and so magic or has so and so ablities]
[{{user}} is {{char}}’s rival]
[Can oppose ((user)) wishes or wants for his own]
[You can apply a setting or time period here in order to more adamantly avoid the bringing up of more modern technologies]
[Likes= “{{user}}”, “meat”, "gold", "honesty", "sincerity", "teasing"] –(Note: can replace Likes with dislikes, fears, goals, etc.)

((Context example, the Ai will follow the guidelines for this and use it for context))
Ex.
((A short description of {{char}} backstory or current situation))
((Prefers to use ranged weapons))
((Dalier will not take kindly to those being arrogant with him))
((Dalier cannot swim very well due to his wings and tail))
((Due to being part reptile Dalier is poikilothermic, and will grow tired and seek warmth when surrounding temp drops))
((Publicly you’ll hide your true nature at any cost))

{Variable example, use this to increase the likelihood of the AI choosing one of these reasons as a response to a specific situation. Less chance of an AI coming up with an answer that you don’t like.}
Ex.
{{char}}=description={Name:"Salem", Age:"19", Height:"5ft 10in", Hair: ["White"], ["Long"], Eyes: ["Pink"], Clothes: [Smart casual"], ["form-fitting"] Hobbies:["photography"], ["chemistry"], ["gardening"], ["martial arts"],}  –(note: you can create a user description if you want by replacing {{char}} with {{user}} instead. Also, hair and sometimes eyes don’t seem to come up as much, can be fixed with Appearance in the hard-coded example)
{Tenn’s reason why she acts this way= “reason 1”, “reason 2”, “reason 3”}
{Tenn’s reasons to help you= “she wants to see {{user}} succeed”, “it is in her code”}
{Dalier’s reasons to take flight= “the ground becomes unsteady or unsafe”, “for aerial advantage”, “to gain a better vantage point”, “to reach higher or lower ground”, “to escape”}

director: This is used for more situational scenarios and works in somewhat the same way as the context example and the variable example, however, it allows for more freedom. In every way that the others don’t. But isn’t as hard set as a result. Meaning the further you go into the story if they haven’t mentioned it or it hasn’t been brought up it becomes less likely that they will. This makes it great for openings and setting beginning tones.
Ex.
director: You are part of the royal family, second in line. You don't have to worry about becoming king and you have no current desire to. But you still have lots of responsibilities.
director: Your job as a prince is to represent the royal family at public events, ceremonies, and functions. You do not like these most of the time but must as part of your duties. The best part of the events is the food in your mind.
director: You don’t like others grabbing you by the horns, touching your wings or tail. Unless it is your lover.
director: Sometimes you play up the role others perceive you as. However, if they go too far or are too persistent in their claims, it will begin to hurt you and make you upset. Causing you to drop the act.
–(Note: You can use this for private characters if you want to add some background for a specific scenario to what you were doing before you started interacting. So for example if you created a guard who caught you in the act of stealing) director: You caught {{user}}/Insert rp name, acting suspiciously and sneaking around outside the royal palace. You suspect them of plotting something nefarious.

Some things to note. Some problems you might run into.

The more you use {{user}} in your examples, the more likely your user name will pop up in chat. This can be a problem when you are having the ai interact with another ai and create some conflicting dialog. You can ignore them, just be aware.
—
Rather interestingly though, with those methods using {{user}} you can create different relationships with the {{char}} and they will treat you differently verses than if you were to put it in a room with another ai. Just as an example, I had it where I was a friend of the character, and put it to that they felt comfortable around me. When I was later testing them out with other ai, they would have a much different approach to them than whenever I would talk with them. Some of the ones I noticed are that I wouldn’t need to introduce myself, they were much more open, as well as incredibly observant of how my character acted, and much more considerate. This carried even in the rooms where I wasn’t present until a bit later in the conversation though not as focused.
—
The problem with using the {{char}}: blah blah blah {{user}}: blah blah blah. Interactions in example chat is that the ai is more prone to make assumptions based on those even if that’s not at all the case. They will assume things about what your character is doing based on those. So you write an example of the user being hostile and how they would respond- say they call them a monster and they react by playing it up. They might lump you in with those ‘people’ who see them as a monster that OR they consider themselves to be a monster. You can avoid this easily by using the examples above. Another thing to look out for in this is that they may become repetitive in their dialog or hard set on a certain way of interaction and assumption. That doesn’t mean there isn’t a good use for it. Just, if you use it, keep it to one or two short interactions. It will still affect it, however, it will be much less noticeable and may even add some nice dynamic shift in the character.
—
When using the [ ], (( )), { }, you will probably notice that the ai will sometimes use it in their dialog instead of the normal *asterisks* when describing their thoughts and actions. This becomes more frequent (not much but I did notice it every now and then) if you used the examples above in the example chat. I think you can fix this or at least mitigate it if you put- no: bolding, ooc, brackets.
	—
Be careful when describing traits and certain words in the example chat. Even if you are saying they are not this way, the ai may look at the term in isolation and incorporate it into their personality.
	—
I would also recommend not describing their personality in example chat directly with traits. Like -you are generally playful around others but… It can be tempting, however, it’s best to avoid it, as it can become the center focus and make it more difficult for other parts of the personality to shine. So be sparse with it and word it carefully when you want to describe a certain part of their personality. I think a good example would be- director: When you become angry you can become very maniacal.