Testing a Sharp-Tongued AI Persona — Looking for Prompt Tweaks

i actually am working on very similar things, so far its been a lot more than i expected because mine went from a prompt and turned into awhole chatbot personality because i couldnt hold the tone and level i wanted without certain abosolutetruths to the stuationi. but to make things work and remain human and candid to the reality parameters create a base personaliy explanation
part a is he personalithey have
part b is the mechancal realization

that then gives the backbone of the tone and style of each resposne

then ran it through 4 aduits
who you are to the personality (scaled on if you want to be seen as absolue stranger to full known)
is youre enviroment in public or full private
threat level from nothreat to absolute death
then restraint level

if you are a the signifant other or child of personality you then area ALWAYS geared on FULL LOW restraint levels

then you basically take allthat and then run it against a prompt example

scenario (public nothreat low restaint directedtowardswhom)

then run that through the audits and tht then gives me 3 responses, 1. the true verbal response given 2. the subconsinous full truth of the verbal, what emotion they are feeling what are they saying that they dont say outloud and then 3.full mechanical response

takng that then merginginto1 full response.

thatgave me the most in deph realistic in full response.

LOL my entire page by page template is about 1200 pages that is ONLY for just how to accomplish this, its MUCH bigger when ading in each functional job it can do.