telegram-bot-builder（Telegram 机器人构建器）

Telegram Bot Builder

Comprehensive guidance for building Telegram bots using the Bot API (v9.4). Covers both Node.js and Python ecosystems with production-ready patterns for authentication, messaging, keyboards, media handling, payments, inline mode, webhooks, and deployment.

When to Use This Skill

Use this skill when:

• Building a new Telegram bot from scratch
• Integrating Telegram messaging into an existing application
• Setting up webhooks or long polling for bot updates
• Creating interactive menus with inline keyboards and callback queries
• Handling media (photos, videos, documents, stickers)
• Implementing Telegram Payments or Telegram Stars
• Building inline mode functionality
• Managing groups, channels, or forum topics via bot
• Deploying bots to production (Docker, PM2, serverless)

Core Concepts

Authentication

Every bot has a unique token obtained from @BotFather. Token format: 123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11.

All API calls go to: https://api.telegram.org/bot<TOKEN>/METHOD_NAME

# .env file
BOT_TOKEN=123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11

Store the token in environment variables. Never commit it to source code.

Receiving Updates: Polling vs Webhook

Long Polling (getUpdates) - Simpler, no HTTPS required, ideal for development:

// Node.js with node-telegram-bot-api
const bot = new TelegramBot(process.env.BOT_TOKEN, { polling: true });

# Python with python-telegram-bot
app = Application.builder().token(os.getenv("BOT_TOKEN")).build()
app.run_polling()

Webhook (setWebhook) - Better for production, lower latency, requires HTTPS (ports 443, 80, 88, or 8443):

bot.setWebHook('https://yourdomain.com/webhook', { secret_token: SECRET });

Choose polling for development and small bots. Choose webhooks for production deployments handling high traffic.

Message Types & Formatting

Send text with sendMessage. Supported parse modes:

• HTML: <b>bold</b>, <i>italic</i>, <code>code</code>, <pre>block</pre>, <a href="url">link</a>, <tg-spoiler>spoiler</tg-spoiler>
• MarkdownV2: *bold*, _italic_, ` code , `` `block` ``, link, ||spoiler||. Requires escaping: _*[]()~>#+-=|{}.!`

Prefer HTML for easier escaping. Use MarkdownV2 when simpler formatting suffices.

Keyboards & Interactive Elements

Inline Keyboard - Buttons attached to messages:

bot.sendMessage(chatId, 'Choose:', {
  reply_markup: {
    inline_keyboard: [
      [{ text: 'Option A', callback_data: 'a' }, { text: 'Option B', callback_data: 'b' }],
      [{ text: 'Visit Site', url: 'https://example.com' }]
    ]
  }
});

Reply Keyboard - Custom keyboard below input field:

bot.sendMessage(chatId, 'Choose:', {
  reply_markup: {
    keyboard: [[{ text: '📊 Stats' }, { text: '⚙️ Settings' }]],
    resize_keyboard: true,
    one_time_keyboard: true
  }
});

Handle inline button presses with callback_query. The callback_data field is limited to 64 bytes. Always call answerCallbackQuery to dismiss the loading indicator.

Sending Media

// Photo (file_id, URL, or upload)
bot.sendPhoto(chatId, 'https://example.com/photo.jpg', { caption: 'A photo' });

// Document
bot.sendDocument(chatId, fs.createReadStream('./file.pdf'), { caption: 'Report' });

// Album (2-10 items)
bot.sendMediaGroup(chatId, [
  { type: 'photo', media: 'https://example.com/1.jpg', caption: 'First' },
  { type: 'photo', media: 'https://example.com/2.jpg' }
]);

Three ways to specify files: file_id (reuse previously uploaded), HTTP URL (Telegram downloads it), or multipart upload. File limits: 50MB upload, 20MB download via Bot API.

Conversation State

For multi-step interactions (registration, forms, wizards), maintain conversation state per chat:

• Node.js: Use a Map or Redis to track {step, data} per chatId
• Python: Use ConversationHandler from python-telegram-bot (built-in state machine)

See reference/patterns_and_examples.md for complete conversation flow implementations.

Error Handling

Handle common error scenarios:

• 429 Too Many Requests: Read retry_after from response, wait, then retry
• 403 Forbidden: Bot was blocked by user or removed from chat
• 400 Bad Request: Invalid parameters (check description field)
• 409 Conflict: Another bot instance using same token with polling

Rate limits: ~30 messages/second to different chats, ~20 messages/minute to same group. Implement exponential backoff for retries.

Bot Commands

Register commands visible in the Telegram menu:

bot.setMyCommands([
  { command: 'start', description: 'Start the bot' },
  { command: 'help', description: 'Show help' },
  { command: 'settings', description: 'Bot settings' }
]);

Commands can be scoped to specific chats, users, or languages using BotCommandScope.

Common Patterns

Quick Start (Node.js)

mkdir my-bot && cd my-bot
npm init -y
npm install node-telegram-bot-api dotenv
echo "BOT_TOKEN=your_token_here" > .env

Quick Start (Python)

mkdir my-bot && cd my-bot
pip install python-telegram-bot python-dotenv
echo "BOT_TOKEN=your_token_here" > .env

Popular Libraries

Key API Method Categories

Deployment Options

• PM2: pm2 start bot.js --name telegram-bot - Process manager with auto-restart
• Docker: Containerized deployment with docker-compose
• Serverless: Webhook handler as Vercel/AWS Lambda function
• VPS: Direct deployment with systemd service

See reference/patterns_and_examples.md for Docker, PM2, and serverless deployment configurations.

Security Checklist

• Store BOT_TOKEN in environment variables
• Validate X-Telegram-Bot-Api-Secret-Token on webhook endpoints
• Verify user IDs for admin commands
• Implement per-user rate limiting
• Sanitize user input before database storage
• Use HTTPS for all webhook endpoints
• Restrict allowed_updates to only needed types

Reference Files

For detailed API documentation and implementation patterns, consult:

• reference/api_methods.md - Complete list of 100+ Bot API methods organized by category (messaging, chat management, stickers, payments, inline mode, games, forum topics, gifts, passport, and more)
• reference/api_types.md - Complete list of 200+ Bot API types with all fields (Update, Message, Chat, User, keyboards, media types, payment types, chat members, reactions, and more)
• reference/patterns_and_examples.md - Production-ready implementation patterns for Node.js and Python including: inline keyboards, webhooks, media handling, conversation state management, database integration, admin panels, multi-language support, Docker/PM2/serverless deployment, Telegram Stars payments, and inline mode

When building a bot, start with SKILL.md for core concepts, then load the appropriate reference file for detailed API information or implementation patterns as needed.