Add AI Chat Bot feature for Kick livestreams

- Introduced a new AI-powered chat bot that interacts with Kick livestream chat.
- Added `transformers` and `torch` as optional dependencies in requirements.txt.
- Created CHAT_FEATURE.md to document the chat bot's features, installation, and usage.
- Implemented chat_service.py to handle AI model loading and message processing.
- Developed kick_chat_bot.py to manage WebSocket connections and chat interactions.
- Added test_chat_bot.py to validate AI chat functionality and integration with Kick.
- Included test_transformers_direct.py for direct testing of transformers library and model loading.
- Established a basic structure for backend services in __init__.py.

Author: H1B0B0

CHAT_FEATURE.md

@@ -0,0 +1,229 @@
+# 🤖 AI Chat Bot for Kick Livestreams
+
+## Overview
+
+KickViewerBOT includes an AI-powered chat bot that can **interact directly with your Kick livestream chat**! The bot reads messages from viewers and responds automatically with AI-generated replies - all running locally on your machine.
+
+## Features
+
+- ✅ **Live Chat Integration** - Connects to Kick chat in real-time
+- ✅ **AI Responses** - Uses Microsoft DialoGPT for natural conversations
+- ✅ **100% Local** - Runs on your PC, no cloud services
+- ✅ **Smart Rate Limiting** - Won't spam chat (configurable response rate)
+- ✅ **Context Aware** - Remembers recent conversation
+- ✅ **Privacy First** - Your conversations never leave your computer
+
+## How It Works
+
+1. **Connects to Kick Chat** - Monitors your stream's chat via WebSocket
+2. **AI Processes Messages** - Lightweight model (117MB) generates responses
+3. **Responds Naturally** - Posts replies back to chat (optional)
+4. **Rate Limited** - Smart timing to avoid spam detection
+
+## Installation
+
+The chat bot requires additional Python packages. Install them with:
+
+```bash
+pip install transformers torch
+```
+
+**Note:** First-time setup will download the AI model (~117MB). This happens automatically on first use.
+
+## Quick Start
+
+### 1. Load the AI Model
+
+The model loads automatically when you start the backend. Check logs for:
+
+```
+🤖 Chat service initialized (model loading in background)
+✅ AI chat model loaded successfully!
+```
+
+### 2. Start Chat Bot for Your Stream
+
+**Via WebSocket:**
+
+```javascript
+socket.emit("kick_chat_start", {
+  channel_name: "your_channel_name",
+  auth_token: "your_kick_auth_token", // Optional - needed to send messages
+  response_chance: 0.2, // 20% chance to respond to each message
+  min_interval: 5, // Minimum 5 seconds between responses
+});
+
+socket.on("kick_chat_started", (data) => {
+  console.log("Bot started:", data);
+});
+```
+
+### 3. Monitor Activity
+
+The bot will:
+
+- 📖 Read all chat messages
+- 🤔 Decide randomly if it should respond (based on `response_chance`)
+- 🤖 Generate AI response with context
+- 💬 Send response to chat (if `auth_token` provided)
+
+## Configuration Options
+
+### Response Chance
+
+- `0.1` = 10% (responds rarely, ~1 in 10 messages)
+- `0.2` = 20% (balanced, recommended)
+- `0.5` = 50% (very active, may seem spammy)
+
+### Min Interval
+
+- `5` seconds = Moderate activity
+- `10` seconds = Conservative
+- `3` seconds = Active (risk of rate limiting)
+
+### Getting Your Auth Token
+
+**To send messages, you need a Kick auth token:**
+
+1. Open Kick.com in browser and login
+2. Open Developer Tools (F12)
+3. Go to Application/Storage → Cookies
+4. Find cookie named `kick_session` or similar
+5. Copy the value
+
+**⚠️ Security Warning:** Never share your auth token publicly!
+
+## How It Works
+
+1. **Model Used:** Microsoft DialoGPT-small
+
+   - Size: 117MB
+   - Type: Conversational AI
+   - Quality: Good for casual chat
+
+2. **Backend Service:** `backend/services/chat_service.py`
+
+   - Loads model asynchronously (doesn't block bot startup)
+   - Maintains conversation history per session
+   - Handles errors gracefully
+
+3. **WebSocket API:**
+   - `chat_status` - Check if chat is ready
+   - `chat_message` - Send message and get response
+   - `chat_clear` - Clear conversation history
+   - `chat_history` - Get full conversation
+
+## Usage
+
+### Check Status
+
+```javascript
+socket.emit("chat_status");
+socket.on("chat_status_response", (data) => {
+  console.log(data);
+  // { available: true, status: 'ready', message: 'AI chat is ready' }
+});
+```
+
+### Send Message
+
+```javascript
+socket.emit("chat_message", {
+  message: "Hello! How are you?",
+  session_id: "user123", // Optional, defaults to socket ID
+});
+
+socket.on("chat_response", (data) => {
+  console.log(data.response);
+  // AI's response
+});
+```
+
+### Clear History
+
+```javascript
+socket.emit("chat_clear", {
+  session_id: "user123",
+});
+```
+
+## Performance
+
+- **First Message:** ~2-5 seconds (model initialization)
+- **Subsequent Messages:** ~0.5-1 second
+- **Memory Usage:** ~500MB RAM when loaded
+- **CPU Usage:** Moderate during generation, idle otherwise
+
+## Troubleshooting
+
+### Model Won't Load
+
+**Error:** `transformers library not installed`
+
+**Solution:**
+
+```bash
+pip install transformers torch
+```
+
+### Slow Response
+
+**Issue:** First response is slow
+
+**Explanation:** Model loads on first use. Subsequent responses are faster.
+
+### Out of Memory
+
+**Issue:** Not enough RAM
+
+**Solutions:**
+
+- Close other applications
+- Restart the bot
+- Use a smaller model (edit `chat_service.py`)
+
+## Alternative Models
+
+You can use different models by editing `backend/services/chat_service.py`:
+
+```python
+# Change this line:
+self.model_name = "microsoft/DialoGPT-small"  # 117MB
+
+# To one of these:
+# self.model_name = "microsoft/DialoGPT-medium"  # 355MB - Better quality
+# self.model_name = "microsoft/DialoGPT-large"   # 774MB - Best quality
+# self.model_name = "distilgpt2"                 # 82MB  - Faster, simpler
+```
+
+## Disabling Chat
+
+If you don't want the chat feature:
+
+1. Simply don't install `transformers` and `torch`
+2. The bot will run normally without chat
+3. Chat endpoints will return "not available" messages
+
+## Security
+
+- **Local Processing:** All AI processing happens on your machine
+- **No Data Collection:** Nothing is sent to external servers
+- **Session Isolation:** Each user has their own conversation context
+
+## Future Improvements
+
+Planned features:
+
+- [ ] Custom system prompts
+- [ ] Model switching via UI
+- [ ] Response streaming for real-time typing effect
+- [ ] Multi-language support
+- [ ] Integration with Kick chat commands
+
+## Need Help?
+
+Open an issue on GitHub with:
+
+- Python version: `python --version`
+- Torch version: `pip show torch`
+- Error logs from the terminal