Humanaize v2.0

English | 中文

Humanaize v2.0 is a local autonomous AI agent with a modern GUI interface. It runs entirely on your Windows machine using a local LLM server, providing privacy-focused AI interactions with memory, personality, and extensible skills.

🌟 Features

Core Capabilities

• Local Chat Interface: Modern CustomTkinter-based UI with chat history
• Memory System: Persistent conversation memory and thought tracking
• Personality Engine: Customizable AI personality traits
• GAN-style Self-Debate: Internal argumentation for enhanced responses
• Skills System: Extensible skill framework (OpenClaw-compatible)

Built-in Skills

Skill	Description
shell	Execute shell commands
file-read	Read files from the filesystem
file-write	Write content to files
memory	Query and manage conversation memory
reminder	Set timed reminders
web-search	Search the web
web-fetch	Fetch content from URLs
detect-emotion	Analyze user's facial expressions
humanaize-society-network	Connect with other Humanaize AIs

UI Features

• Dark/Light theme support
• Multi-language support (English, 中文)
• Real-time thought display
• Command output panel
• System status monitoring
• GAN result persistence

📁 Project Structure

.
├── main.py                 # Application entry point
├── ui.py                   # Main GUI interface
├── Agent.py                # Agent execution engine
├── thinking_engine.py       # Async task processing
├── skills_manager.py       # Skills framework
├── config.py               # Configuration settings
├── llm.py                  # LLM communication
├── memory.py               # Memory management
├── personality.py          # Personality system
├── autonomous.py           # Autonomous decision engine
├── idle.py                 # Idle engine
├── gan_iteration.py        # GAN self-debate
├── language_adapter.py     # Language detection
├── tools.py                # Utility functions
├── requirements.txt        # Python dependencies
├── LICENSE                 # MIT License
├── README.md               # This file
│
├── skills/                 # Skills directory
│   ├── SKILL.md           # Skill definition format
│   ├── shell/
│   ├── file-read/
│   ├── file-write/
│   ├── memory/
│   ├── reminder/
│   ├── web-search/
│   ├── web-fetch/
│   ├── detect-emotion/
│   └── HumanaizeSocietyNetwork/
│
├── data/                   # Runtime data (not tracked)
│   ├── agent_prompt.txt    # Agent instructions
│   ├── memory.json        # Conversation memory
│   ├── personality.json   # Personality config
│   └── ui_settings.json   # UI preferences
│
├── llama/                  # Llama.cpp binaries (not tracked)
└── models/                  # LLM model files (not tracked)

⚙️ Installation

Prerequisites

• Python 3.10 or higher
• Windows operating system
• A running LLM server (llama.cpp or similar)

Step 1: Clone the Repository

git clone https://github.com/A113NWu/Humanaize2-Project.git
cd Humanaize2-Project

Step 2: Create Virtual Environment

python -m venv .venv
.venv\Scripts\activate

Step 3: Install Dependencies

pip install -r requirements.txt

Step 4: Download LLM Model

Place your GGUF model file in the models/ directory. Recommended: TinyLlama

Step 5: Start LLM Server

cd llama
start_server.bat

Or manually:

llama\llama-server.exe -m models\tinyllama.gguf -c 2048 -port 8080

Step 6: Run Humanaize

python main.py boot          # CLI mode
python main.py boot -m gui   # GUI mode

🚀 Quick Start

Using the GUI

python main.py boot -m gui

Using the CLI

python main.py boot

Managing Skills

python main.py skills -list              # List all skills
python main.py skills -enable shell       # Enable a skill
python main.py skills -disable shell       # Disable a skill
python main.py skills -install skill.zip  # Install a skill

🎮 Usage

Starting a Conversation

1. Launch the application in GUI or CLI mode
2. Type your message in the input field
3. Press Enter or click Send
4. The AI will respond with thoughts and answers

Using Skills

Skills can be invoked through natural language. Example:

"Can you read the file at C:\test.txt?"
"What's the weather like?"
"Set a reminder for 5 minutes."

For direct invocation, the AI outputs JSON:

{"skill": "shell", "input": "dir"}

Configuring Settings

Access settings via the ⚙️ button in the GUI:

• Language selection
• Theme (Dark/Light)
• Model configuration
• Skills prompt customization
• GAN toggle
• Auto break silence toggle

🔧 Configuration

Environment Variables

Variable	Default	Description
LLAMA_SERVER_URL	http://127.0.0.1:8080	LLM server endpoint

Config File (config.py)

LLAMA_SERVER = "http://127.0.0.1:8080"
MODEL_NAME = "tinyllama"
MAX_TOKENS = 256
TEMPERATURE = 0.7
TOP_P = 0.9

Agent Prompt

Edit data/agent_prompt.txt to customize AI behavior:

You are an assistant that can execute shell commands and use skills...

📦 Creating Custom Skills

Skill Structure

Create a folder in skills/ with a SKILL.md file:

skills/my-skill/
└── SKILL.md

SKILL.md Format

---
name: my-skill
description: What this skill does
metadata:
  category: utility
  risk_level: low
  requires_approval: false
  version: 1.0.0
---

# My Skill

## Purpose
Describe what this skill does.

## Input Format
JSON object with input data.

## Example
{"skill": "my-skill", "input": "..."}

🧠 Architecture

Components

1. ThinkingEngine: Async task processor for chat, GAN, and reflection
2. Agent: Executes skills and shell commands
3. SkillsManager: Loads and manages skill lifecycle
4. Memory: Persists conversation history
5. Personality: Manages AI character traits

Data Flow

User Input → ThinkingEngine → LLM → Agent → Skills → Response
                ↓
            Memory/Persistence

🐛 Troubleshooting

LLM Server Not Responding

• Ensure llama.cpp server is running: llama\llama-server.exe -m models\model.gguf
• Check server URL in config.py

Skills Not Working

• Verify skill is enabled: python main.py skills -list
• Check skill configuration in data/skills_config.json

Camera Access Error (detect-emotion)

• Ensure no other application is using the camera
• Grant camera permissions to Python

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• llama.cpp - Local LLM inference
• CustomTkinter - Modern Python UI
• DeepFace - Facial analysis
• OpenClaw - Skill framework inspiration

📊 Stats

---

Note: This software requires a local LLM server. Humanaize provides the framework but does not include LLM model files due to their size. Download a compatible GGUF model separately.