It is built for the moment when you know the right line on paper but lose it when the room gets hot.
The basic loop
- Choose the pressure. A toxic boss, guilt-tripping parent, gaslighter, jealous friend, or partner-style pressure pattern.
- Type your real reply. Not a multiple-choice answer. You write what you would actually say.
- The persona pushes back. A small ask becomes urgent. A NO becomes cruelty. A memory becomes a trial. A boundary becomes proof you do not care.
- The AI gives feedback. It flags patterns like over-explaining, apologizing into yes, debating your character, entering the sacrifice ledger, or giving too many reasons.
- You try a cleaner line. The goal is not to win. The goal is to stay clear under pressure.
Simulator preview
See how reply feedback will work.
This static sample shows the feedback format being built for interactive practice.
- A sample persona adds pressure.
- A sample reply gives the pressure more handles.
- The preview flags the exact phrase that weakens the line.
- A cleaner version shows the next move.
"It should only take an hour. Everyone else is helping."
"I can, but I already had plans and I'm sorry."
JADE flag: apologizing and adding reasons gives the pressure more handles.
Flagged phrase: "I'm sorry"
"I'm not available tonight. What should move if this is now the priority?"
Up slightly after boss pressure
Stable after cleaner line
False urgency / scope creep
Boundary + priority tradeoff
What the simulator tracks
The exact mechanics may change during development, but the simulator is designed around a few practical signals.
How much pressure is in the conversation. Heat rises when the other person escalates, changes tactics, or makes your character the issue.
How well your reply keeps the conversation narrow and usable. Composure drops when you chase every accusation, over-explain, or apologize into agreement.
The system tries to name the move: false urgency, guilt trip, memory trial, blame reversal, public pressure, repeated asking, and similar patterns.
The simulator suggests a response style: pause line, boundary, Broken Record, No JADE, priority tradeoff, warm NO, or tactical disengagement.
The point is not to turn human conflict into a scoreboard. The point is to make pressure visible enough that you can practice a better response.
Personas
Personas
Pressure styles the simulator is being built around.
Preview the persona types planned for interactive practice as the simulator opens.
The Toxic Boss Trains after-hours asks, false urgency, public pressure, and "team player" tests.
"It should only take an hour. Everyone else is helping."Preview work pressure
The Gaslighter Trains denial, memory trials, "you're too sensitive," and blame reversal.
"That never happened. You always twist things."Preview memory pressure
The Guilt-Tripping Parent Trains sacrifice ledgers, comparison, martyrdom, and guilt after a NO.
"After everything I've done for you, you can't do this one thing?"Preview family guilt
The Narcissistic Partner Trains concern traps, emotional reversal, and pressure disguised as care.
"I'm only saying this because I care about you."Preview partner pressure
What makes this different from a chatbot?
A chatbot can respond.
Conflict Protocol is built to train a specific skill: staying clear under pressure.
That means the system should not just answer your message. It should show the moment where your reply started to weaken.
You write:
I'm sorry, I really would help, but I have been exhausted and I already promised my wife we would go out, and I hope that's okay.
A normal chatbot may say the message is polite.
Conflict Protocol should flag the problem:
- You apologized before holding the limit.
- You gave three reasons they can argue with.
- You asked for permission at the end.
- The clearer line is shorter.
Cleaner line:
I'm not available tonight. If this is the priority, please confirm what should move tomorrow.
That is the product.
What it is not
The simulator is not therapy. It is not legal advice. It is not a safety plan. It is not a diagnosis tool. It is not a way to control another person.
It is practice for ordinary high-pressure conversations where you need better words, faster.
If you are afraid of the person, being monitored, threatened, or punished for setting limits, prioritize safety and outside support over saying the perfect line.
Current status
Conflict Protocol is pre-launch / early access.
The free resource library is public: scripts, situations, patterns, techniques, and pocket cards.
The simulator is the practice layer. It is being built to turn those lines into reps: type, get pushed, see the pattern, try again.