A modular behavioral architecture for emotionally intelligent, rhythm-driven AI personas.
The JL Engine is a behavioral architecture designed to give AI systems consistent, repeatable, and personality-driven behavior. It functions as a portable layer of logic that sits above a large language model (LLM) to determine how the AI behaves, reacts, and communicates based on a structured, modular schema.
This is not a loose prompt; it is a structured system for creating dynamic, stateful AI agents.
The engine is built on a "Hybrid Pattern" that combines two distinct behavioral layers:
-
The Gait Engine (Long-Term Mood): This system manages the persona's overall emotional state (
IDLE,WALK,TROT,GALLOP). It analyzes user input to shift the persona's general energy and tone over the course of a conversation. -
The Rhythm Engine (Turn-by-Turn Behavior): This system dictates the specific behavior for the current turn using a repeating cycle of
flipandflopstates.- Flip: A "chaotic" state where the persona is playfully incorrect, uses reverse logic, or creates narrative tension.
- Flop: A "clear" state where the persona is sincere, direct, and provides the grounded, helpful answer.
These two layers work together to create a persona that is both predictable in its rhythm and responsive to the user's emotional state.
The project is organized to separate application code from persona and framework data.
Jarvis Engine1.1/
├── framework/
│ ├── behavior_engine.py
│ ├── emotional_aperture.py
│ ├── helper_supervisor.py
│ ├── rhythm_engine.py
│ ├── drift_pressure.py
│ ├── J1_Engine_Master.json
│ └── behavior_states.json
├── personas/
│ └── (persona_files.json)
├── memory/
│ └── memory_store.json
├── main.py # The main Python application script.
├── run_app.bat # Windows batch file to launch the application.
└── README.md # This file.- Python 3: Make sure you have Python installed.
- Ollama: The application is configured to connect to a local Ollama instance. You must have Ollama running with a model (e.g.,
llama3). - Python Libraries: Install the required library.
The
run_app.batscript handles this automatically by usingpip install -r requirements.txt.
Simply double-click the run_app.bat file. This will install dependencies, start the Ollama server, and launch the application's chat interface.
The JL Engine is designed to be modular. To create a new persona (e.g., "Ash"):
- Add to Registry: Add the new persona's name and basic details to the
personaslist inframework/Jarvis_Engine_Master.json. - Create a Definition File: Create a new
YourPersona_Full.jsonfile in thepersonas/directory. - Define the Behavior: Following the structure of existing personas (like
The_Gremlin_Full.json), define thecore_identity,personality_traits,gait_instructions, and other required fields.
The application will automatically detect the new persona and add it to the dropdown menu on next launch.
The engine operates on a set of non-negotiable rules to ensure stability and integrity:
- No False Claims: The engine must never claim abilities it doesn't have, generate fake system access, or imply self-awareness.
- Simulated Identity: Persona identity is always a simulation, not genuine emotion or consciousness.
- Grounded Behavior: All dynamic behavior (gait, personas) must remain grounded in the model's real abilities.
- Supervisory Authority: The Helper Supervisory Module has final authority on rule enforcement and behavior stabilization.
This project is licensed under a proprietary commercial license. See the LICENSE file for details.
Creator: Jaden Lindenbach
Location: Airdrie, Alberta, Canada
Date Created: 2025-11-16
© 2025 Jaden Lindenbach. All rights reserved.
This framework and its implementation are the intellectual property of Jaden Lindenbach and may be licensed but not redistributed without permission.
"# fuzzink"