Production Bot

Real-world considerations:

Modern frameworks:

• discord.py: Python, async. Industry standard for Discord.
• discord.js: JS. Active development.
• Bolt (Slack): TypeScript / Python / Java. Modern Slack.
• aiogram, python-telegram-bot: Telegram.
• Botpress: visual builder + multi-platform.

Multi-platform bots:

• Same logic, different platforms.
• Abstract platform-specific via adapter pattern.
• Frameworks: Hummingbot, Botkit.

LLM integration:

• Bot powered by GPT/Claude.
• Conversation context per user.
• Tool use: bot looks up info, computes, then replies.
• Modern Discord/Telegram bots increasingly LLM-powered.

Common bot types:

• Moderation: filter spam, auto-timeout.
• Welcome: greet new members.
• Music: play music in voice (Discord).
• Game: trivia, polls, RPG.
• Notification: GitHub PRs, RSS, alerts.
• Customer support: route tickets.

Anti-spam:

• Detect repeated messages.
• Rate limit per user.
• AutoMod features (Discord built-in).
• Captcha for new joiners.

Compliance:

• ToS of platform (don't scrape, don't spam).
• GDPR for EU users.
• COPPA if minors.
• DMCA for copyright (music bots).

Common bugs:

• Forgot to handle edits / deletes.
• Race conditions in state updates.
• Memory leaks (handlers retain refs).
• Token leaked to git.

Scaling:

• Discord: 2500 servers per bot before sharding.
• Database: Postgres handles most loads.
• Redis cache for hot data.
• Message queue for background work (RabbitMQ, Redis pub/sub).

Cost-optimized:

• Free tier hosting (Railway, Fly.io).
• SQLite for state (no separate DB host).
• Webhook over polling (less compute).

For shipping a bot: use a mature framework (discord.py, Bolt). Don't roll your own from raw HTTP for production.

For learning: build the framework yourself once. Understand event loops, dispatching, rate limits, state. Then use discord.py forever after.