Telegram Bot Builder

Overview

Builds production-ready Telegram bots covering the full Bot API surface: commands, inline keyboards, callback queries, conversations with state machines, media handling, group management, payments, and Mini Apps. Supports both long polling (development) and webhooks (production).

Instructions

1. Bot Creation

1. Message @BotFather on Telegram
2. Send /newbot, choose name and username
3. Save the bot token (format: 123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11)
4. Configure with /setcommands, /setdescription, /setabouttext

2. Project Scaffolding

Recommended frameworks by language:

Node.js — grammY (recommended):

npm init -y && npm install grammy @grammyjs/conversations

Node 20.6+ loads env files natively: node --env-file=.env bot.js (no dotenv needed).

Node.js — Telegraf (4.16.3 is its latest release; it lags behind newer Bot API features, so prefer grammY for new bots):

npm install telegraf

Python — python-telegram-bot:

pip install python-telegram-bot       # v22, async API, Python 3.10+

# bot.py
import os
from telegram import Update
from telegram.ext import Application, CommandHandler, ContextTypes

async def start(update: Update, context: ContextTypes.DEFAULT_TYPE):
    await update.message.reply_text("Welcome!")

app = Application.builder().token(os.environ["BOT_TOKEN"]).build()
app.add_handler(CommandHandler("start", start))
app.run_polling()

Standard project structure:

telegram-bot/
├── bot.js                # Entry point, bot initialization
├── handlers/
│   ├── commands.js       # /start, /help, custom commands
│   ├── callbacks.js      # Inline keyboard callback handlers
│   ├── conversations.js  # Multi-step conversation flows
│   └── middleware.js      # Auth, logging, rate limiting
├── keyboards/            # Inline and reply keyboard builders
├── services/             # Business logic
├── .env                  # BOT_TOKEN, WEBHOOK_URL, WEBHOOK_SECRET, ADMIN_ID (gitignored)
└── package.json

3. Core Patterns (grammY)

Bot setup (ESM, "type": "module" in package.json):

import { Bot, InlineKeyboard, InputFile, session, webhookCallback } from 'grammy';
import { conversations, createConversation } from '@grammyjs/conversations';

const bot = new Bot(process.env.BOT_TOKEN);
bot.catch((err) => console.error('Update failed:', err.error));

Basic command handler:

bot.command('start', async (ctx) => {
  await ctx.reply('Welcome!', {
    reply_markup: {
      inline_keyboard: [[
        { text: '📊 Dashboard', callback_data: 'dashboard' },
        { text: '⚙️ Settings', callback_data: 'settings' }
      ]]
    }
  });
});

Callback query handler:

bot.callbackQuery('dashboard', async (ctx) => {
  await ctx.answerCallbackQuery();
  await ctx.editMessageText('Here is your dashboard...', {
    reply_markup: backButton
  });
});

Conversation (conversations plugin v2; no session needed):

async function onboarding(conversation, ctx) {
  await ctx.reply('What is your name?');
  const name = await conversation.waitFor('message:text');
  await ctx.reply('What is your email?');
  const email = await conversation.waitFor('message:text');
  await ctx.reply(`Thanks ${name.message.text}! Registered with ${email.message.text}`);
}

bot.use(conversations());
bot.use(createConversation(onboarding, 'onboarding'));   // function first, name second
bot.command('register', (ctx) => ctx.conversation.enter('onboarding'));

Register conversations() after any session() middleware and before the conversation. Code in a conversation is replayed, so wrap side effects (database writes, API calls) in conversation.external(...).

Middleware for auth:

function adminOnly(ctx, next) {
  if (ctx.from?.id !== Number(process.env.ADMIN_ID)) {
    return ctx.reply('⛔ Not authorized');
  }
  return next();
}
bot.command('admin', adminOnly, handleAdmin);   // grammY: bot.command(name, ...middleware)

4. Keyboard Types

Inline Keyboard (attached to message):

• Callback buttons (callback_data) — triggers callbackQuery handler
• URL buttons (url) — opens a link
• Web App buttons (web_app) — opens a Mini App
• Switch Inline buttons (switch_inline_query) — triggers inline mode

Reply Keyboard (replaces phone keyboard):

• Custom keyboard with predefined responses
• one_time_keyboard: true to auto-hide after selection
• resize_keyboard: true for compact display

Remove Keyboard:

{ reply_markup: { remove_keyboard: true } }

5. Polling vs Webhooks

Long Polling (development):

bot.start(); // Calls getUpdates in a loop

• No public URL needed
• Slightly higher latency
• Good for development and low-traffic bots

Webhooks (production):

import express from 'express';
const app = express();
app.use(express.json());
// A secret header proves the request comes from Telegram; keep the token out of the URL
app.post('/telegram/webhook', webhookCallback(bot, 'express', { secretToken: process.env.WEBHOOK_SECRET }));
app.listen(3000);

await bot.api.setWebhook('https://bots.northwind.dev/telegram/webhook', {
  secret_token: process.env.WEBHOOK_SECRET,
});

• Requires a public HTTPS URL on port 443, 80, 88 or 8443, with a valid (non-wildcard) certificate and no redirects
• Answer within about 10 seconds or Telegram re-sends the update; push slow work to a queue
• Webhook and getUpdates are mutually exclusive: call bot.api.deleteWebhook() before going back to polling

6. Media Handling

// Send photo
await ctx.replyWithPhoto(new InputFile('./image.png'), { caption: 'Check this out' });

// Send document
await ctx.replyWithDocument(new InputFile(buffer, 'report.pdf'));

// Handle received photos
bot.on('message:photo', async (ctx) => {
  const file = await ctx.getFile();
  const url = `https://api.telegram.org/file/bot${process.env.BOT_TOKEN}/${file.file_path}`;
});

7. Bot API Limits

Current Bot API is 10.x (10.3, August 2026); check the changelog at core.telegram.org/bots/api-changelog for new features.

• Messages: about 1 per second per chat, 20 per minute in one group, about 30 per second for broadcasts overall (paid broadcasts allow up to 1000/s, charged in Stars). Over the limit gives HTTP 429 with retry_after; honor it (grammY: @grammyjs/auto-retry)
• Inline results: 50 per query
• File upload: 50 MB per file through the cloud Bot API (a self-hosted Bot API server raises this)
• File download: 20 MB via getFile
• Message length: 4096 characters; caption: 1024 characters
• Inline keyboard: 100 buttons per message
• Callback data: 1-64 bytes

8. Deployment

• Simple: Fly.io, Render, Railway or any container host (webhook mode)
• Serverless: Vercel/AWS Lambda with webhook adapter
• VPS: systemd service with auto-restart
• Docker: Lightweight Node.js container with health checks

Examples

Example 1: Task Management Bot

Input: "Build a Telegram bot for my team to manage tasks. Users should be able to create tasks, assign them, set deadlines, and get reminders."

Output: A grammY bot with:

• /newtask command opening a conversation flow (title → description → assignee → deadline)
• Inline keyboard for task status updates (To Do → In Progress → Done)
• Daily reminder messages for overdue tasks using node-cron
• /mytasks showing personal task list with inline navigation
• SQLite database for persistence via better-sqlite3

Example 2: Content Publishing Bot

Input: "Create a Telegram bot that lets me draft posts, preview them, schedule them to a channel, and track engagement."

Output: A grammY bot with:

• Multi-step drafting flow with text, photos, and formatting preview
• Schedule picker with inline calendar keyboard
• Auto-publishing to target channel via bot.api.sendMessage(channelId, ...)
• Post links with a per-post tracking parameter (the Bot API does not expose channel view counts to bots)
• /drafts command listing scheduled posts with edit/delete options

Guidelines

• Always register bot.catch(...) — unhandled errors in handlers stop the bot
• Use ctx.answerCallbackQuery() to dismiss the loading indicator on buttons
• Store the bot token in env vars, never hardcode or log it; revoke a leaked one with /revoke in @BotFather
• Payments for digital goods use Telegram Stars (currency XTR); answer pre_checkout_query within 10 seconds
• Use parse_mode: 'HTML' or 'MarkdownV2' for rich text (MarkdownV2 requires escaping special chars)
• Implement graceful shutdown: bot.stop() on SIGINT/SIGTERM
• For groups: handle my_chat_member updates to track when bot is added/removed
• Set commands list via bot.api.setMyCommands() for autocomplete
• Show progress with await ctx.replyWithChatAction('typing') (or the auto-chat-action plugin) for long operations
• Rate limit user interactions to prevent abuse
• For conversations: always handle the case where the user sends an unexpected message type