NexChat - AI Chat Assistant

English | 简体中文

A feature-rich AI chat assistant built with Streamlit, supporting multiple AI models, role-playing mode, and conversation management.

Features

• 🤖 Multiple AI Models: Support for various AI models with customizable API endpoints
• 👤 User Management: User registration and login system with conversation history
• 🎭 Role-Playing Mode: Engage in immersive role-playing conversations with predefined or custom characters
• 💬 Normal Chat Mode: Standard AI conversation mode
• 📊 Conversation Management: Save and manage chat history in SQLite database
• 👍👎 Message Feedback: Like/dislike feedback system for AI responses
• ⚙️ Customizable Settings: Adjustable temperature, max tokens, timeout, and retry settings

Installation

Prerequisites

• Python 3.8 or higher
• pip package manager

Setup

1. Clone the repository:

git clone https://github.com/nex-agi/NexChat.git
cd NexChat

2. Install required dependencies:

pip install -r requirements.txt

3. Configure your models in config.yml (see Configuration section)

4. Run the application:

streamlit run app.py

Configuration

The application uses config.yml for configuration. All model settings, API keys, and application preferences are configured in this file.

Configuration Structure

# Application Configuration
app:
  title: "AI Chat Assistant"
  page_icon: "🤖"
  layout: "wide"
  initial_sidebar_state: "expanded"

# UI Configuration
ui:
  show_model_info: true  # Whether to show model information
  show_token_count: true  # Whether to show token count

# Available Models List
models:
  - name: "model-name"
    display_name: "Model Name"
    description: "Model Description"
    api_key: "your-api-key-here"
    base_url: "https://api.example.com/v1"  # Replace with actual base URL

Adding Models

To add a new model, add an entry to the models list in config.yml:

models:
  - name: "your-model-name"
    display_name: "Your Model Display Name"
    description: "Model description"
    api_key: "your-api-key"
    base_url: "https://your-api-endpoint.com/v1"

Usage

Getting Started

1. Login/Register: Enter your username in the sidebar and click "Login/Register"
2. Select Model: Choose an AI model from the model selector
3. Choose Chat Mode:
   • Normal Chat: Standard AI conversation
   • Role-Playing: Engage with predefined or custom characters
4. Start Chatting: Type your message in the input box and press Enter

Role-Playing Mode

1. Select "Role-Playing" from the chat mode options
2. Choose a character from the dropdown or create a custom character
3. Custom characters can be added with:
   • Character name
   • Character description (background, personality, speaking style, etc.)
4. Start conversing with your selected character

Features

• Message Feedback: Click 👍 or 👎 on AI responses to provide feedback
• Clear Chat: Use the "Clear Chat History" button to start a new conversation
• Settings Adjustment: Modify temperature, max tokens, timeout, and retry settings in the sidebar
• Logout: Click "Logout" to save your current conversation and exit

Project Structure

NexChat/
├── app.py              # Main application file
├── config.yml          # Configuration file
├── prompt.md           # Role-playing prompt template
├── char.jsonl          # Predefined character data (JSONL format)
├── chat_data.db        # SQLite database for conversation storage
├── requirements.txt    # Python dependencies
└── README.md           # This file

Database Schema

The application uses SQLite to store:

• users: User information (id, username, created_at)
• conversations: Chat messages (id, user_id, session_id, role, content, timestamp, model)
• message_feedback: User feedback on messages (id, user_id, session_id, message_index, feedback_type, timestamp)

Notes

• The application automatically creates the SQLite database on first run
• Character data should be in JSONL format in char.jsonl
• Role-playing prompts are loaded from prompt.md
• API keys should be kept secure and not committed to version control
• Some models may require proxy configuration (see code for specific model handling)