replyt

Replyt

Replyt is a TypeScript npm package that allows you to automate YouTube comment replies using AI. Built with OpenAI GPT and YouTube Data API v3.

🚀 Features

• ✅ Full Automation: Automatically reply to YouTube comments
• 🤖 Smart AI: Uses OpenAI GPT to generate natural responses
• 🌍 Multi-language: Replies in the same language as the comment
• 📊 Database Tracking: Tracks processed comments
• ⚡ Rate Limiting: Respects YouTube API limits
• 🔧 Configurable: Customize reply style, intervals, and more
• 📦 CLI & Library: Use as command line tool or import as library

📦 Installation & Usage

There are two main ways to use Replyt:

🎯 Method 1: CLI Tool (Global Installation)

Best for: Quick setup, production use, non-developers

# Install globally
npm install -g replyt

# Setup configuration
replyt setup

# Run the bot
replyt

🛠️ Method 2: Local Development (Clone Repository)

Best for: Developers, customization, contributing, learning

# Clone repository
git clone https://github.com/rfqma/replyt.git
cd replyt

# Install dependencies
npm install

# Setup configuration
npm run setup

# Run in development mode
npm run dev

🔧 Setup Guide

Environment Variables

Both methods require a .env file with your API credentials:

# YouTube API
YOUTUBE_API_KEY=your_youtube_api_key_here
YOUTUBE_CHANNEL_ID=your_channel_id_here

# OpenAI
OPENAI_API_KEY=your_openai_api_key_here

# OAuth (optional - for posting comments)
YOUTUBE_CLIENT_ID=your_client_id_here
YOUTUBE_CLIENT_SECRET=your_client_secret_here
YOUTUBE_REFRESH_TOKEN=your_refresh_token_here
YOUTUBE_ACCESS_TOKEN=your_access_token_here

# Bot Configuration
CHECK_INTERVAL_MINUTES=5
MAX_REPLIES_PER_RUN=10
REPLY_STYLE="friendly and helpful"
DATABASE_PATH=./data/replyt.db

Setup APIs

For CLI (Global Install):

replyt setup    # Interactive setup wizard

For Local Development:

npm run setup   # Setup YouTube API Key
npm run oauth   # Setup OAuth for posting (optional)
npm run test    # Test configuration

🎯 Usage Instructions

🌐 CLI Usage (Global Install)

After global installation and setup:

# Start the bot
replyt

# That's it! The bot will run automatically

💻 Local Development Usage

Running the Bot:

npm run dev     # Development mode with auto-restart
npm run start   # Production mode (requires build)
npm run watch   # Development with file watching

Development Commands:

npm run build   # Compile TypeScript
npm run clean   # Clean build files
npm run test    # Test API connections
npm run setup   # Run setup wizard
npm run oauth   # Setup OAuth authentication

📚 Library Usage (Both Methods)

You can also use Replyt as a library in your own projects:

# Install as dependency
npm install replyt

import { Replyt, AutoReplyBot, config } from "replyt";

// Using the Replyt wrapper class
const bot = new Replyt(config);

// Start bot with cron scheduling
await bot.start("*/5 * * * *"); // Every 5 minutes

// Or use manually
await bot.initialize();
await bot.processNewComments();
const stats = await bot.getStats();

Advanced Usage with AutoReplyBot

import { AutoReplyBot, config, validateConfig } from "replyt";

// Validate configuration
validateConfig();

// Create bot instance
const bot = new AutoReplyBot(config);

// Initialize
await bot.initialize();

// Process comments once
await bot.processNewComments();

// View statistics
const stats = await bot.getStats();
console.log(stats);

// Shutdown
bot.shutdown();

📋 API Documentation

Classes

Replyt

Wrapper class for easy library usage.

class Replyt {
  constructor(config: Config);
  async initialize(): Promise<void>;
  async processNewComments(): Promise<void>;
  async getStats(): Promise<object>;
  async start(cronExpression?: string): Promise<void>;
  shutdown(): void;
}

AutoReplyBot

Core bot class with full control.

class AutoReplyBot {
  constructor(config: Config);
  async initialize(): Promise<void>;
  async processNewComments(): Promise<void>;
  async getStats(): Promise<object>;
  shutdown(): void;
}

Services

YouTubeService

• Fetch comments from channel
• Post replies to YouTube
• OAuth management

OpenAIService

• Generate AI responses
• Filter comments that need replies
• Multi-language support

DatabaseService

• SQLite database for tracking
• Store comment status
• Processing statistics

Types

interface Config {
  youtubeApiKey: string;
  youtubeChannelId: string;
  youtubeOAuth: OAuthCredentials;
  openaiApiKey: string;
  checkIntervalMinutes: number;
  maxRepliesPerRun: number;
  replyStyle: string;
  databasePath: string;
}

interface YouTubeComment {
  id: string;
  videoId: string;
  authorDisplayName: string;
  textDisplay: string;
  publishedAt: string;
  // ... other properties
}

⚙️ Configuration

Environment Variables

Variable	Required	Default	Description
YOUTUBE_API_KEY	✅	-	YouTube Data API v3 key
YOUTUBE_CHANNEL_ID	✅	-	Your YouTube channel ID
OPENAI_API_KEY	✅	-	OpenAI API key
YOUTUBE_CLIENT_ID	❌	-	OAuth client ID (for posting)
YOUTUBE_CLIENT_SECRET	❌	-	OAuth client secret
YOUTUBE_REFRESH_TOKEN	❌	-	OAuth refresh token
CHECK_INTERVAL_MINUTES	❌	5	Comment check interval (minutes)
MAX_REPLIES_PER_RUN	❌	10	Max replies per cycle
REPLY_STYLE	❌	"friendly and helpful"	AI reply style
DATABASE_PATH	❌	"./data/replyt.db"	SQLite database path

Reply Styles

You can customize AI reply style by changing REPLY_STYLE:

• "friendly and helpful"
• "professional and formal"
• "casual and funny"
• "educational and informative"
• Custom: "Always reply in Indonesian with emojis"

🔒 Read-Only Mode

If you don't configure OAuth, the bot will run in read-only mode:

• ✅ Read and analyze comments
• ✅ Generate AI responses
• ✅ Track in database
• ❌ Won't post replies to YouTube

To enable posting:

CLI: replyt oauth Local: npm run oauth

📊 Monitoring

The bot provides detailed logging and statistics:

🤖 Bot initialized successfully - FULL MODE (can post replies)
📝 Found 25 total comments
🆕 Found 5 new comments to process
🧠 Generating reply for comment by John Doe...
📤 Posting reply: "Thank you for watching! ..."
✅ Successfully replied to comment abc123
📊 Total comments processed: 150

🛠️ Development (Local Repository)

Quick Start

git clone https://github.com/rfqma/replyt.git
cd replyt
npm install
npm run setup
npm run dev

Available Scripts

Command	Description
npm run dev	Development mode with auto-restart
npm run start	Production mode (requires build)
npm run build	Compile TypeScript
npm run watch	Development with file watching
npm run setup	Setup wizard
npm run oauth	OAuth setup
npm run test	Test configuration
npm run clean	Clean build files
npm run validate	Validate package structure

Project Structure

replyt/
├── src/
│   ├── bin/replyt.ts      # CLI executable
│   ├── lib/index.ts       # Library exports
│   ├── services/          # Core services
│   ├── config/            # Configuration
│   └── types/             # TypeScript types
├── scripts/               # Setup scripts
├── examples/              # Usage examples
└── dist/                  # Compiled output

🔧 Troubleshooting

Common Issues

• API Key Invalid

  Error: The request cannot be completed because you have exceeded your quota.
  

  • Check YouTube API key and quota
  • Ensure YouTube Data API v3 is enabled

• OAuth Issues

  Error: invalid_grant
  

  • Refresh token expired, run OAuth setup again:
    • CLI: replyt oauth
    • Local: npm run oauth

• OpenAI Rate Limits

  Error: Rate limit exceeded
  

  • Reduce MAX_REPLIES_PER_RUN
  • Increase CHECK_INTERVAL_MINUTES

Debug Mode

CLI: DEBUG=replyt:* replyt Local: DEBUG=replyt:* npm run dev

📄 License

MIT License - see LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please:

• Fork the repository
• Create a feature branch (git checkout -b feature/AmazingFeature)
• Commit your changes (git commit -m 'Add some AmazingFeature')
• Push to the branch (git push origin feature/AmazingFeature)
• Open a Pull Request

📞 Support

• 🐛 Report bugs
• 💡 Request features
• 📧 [e-mail]

🙏 Credits

Built with ❤️ using:

• OpenAI GPT for AI responses
• YouTube Data API v3 for YouTube integration
• TypeScript for type safety
• SQLite for data persistence

⭐ Star this repository if it's helpful!