Telegram Group Moderation Assistant

• Someone posts a message in your group or supergroup
• The bot checks it currently has the permissions it needs to act (delete, mute, ban) before doing anything
• AI reads the message against your moderation policy and decides on exactly one outcome: allow, warn, delete, mute, or ban
• For a warning, the bot posts a public reminder in the chat; for delete, mute or ban it acts directly through Telegram — no manual step in between
• Nothing happens for messages that are fine — they're allowed through without any trace that the bot even looked at them

• Add the bot to your group or supergroup.
• Promote it to administrator and grant "Delete messages" plus "Ban/restrict members" — without these, warnings still work but delete, mute and ban actions will fail.
• Disable Privacy Mode for the bot in BotFather so it actually sees every message, not just ones that mention it.
• Before turning it loose on a live community, review the moderation policy and the AI prompt, and run it in observe-only mode for a while to see what it would have done.
• Once you trust the calls it's making, switch from observe-only to delete-and-warn or full auto-enforcement.

Good to know

• The bot needs to be a group administrator with "Delete messages" and "Restrict/ban members" permissions before enforcement can actually happen — it checks its own permissions before every decision, but a missing permission means the delete/mute/ban step will fail rather than silently succeed.
• It ships watching every message by default, which means you also need to disable Privacy Mode for the bot in BotFather — otherwise Telegram only forwards messages that mention it, and most spam and abuse won't be visible at all.
• The moderation policy and the AI's instructions are both editable in the visual editor, so you can tune exactly what counts as spam, abuse or off-topic for your community.
• This is a starter kit built for the observe-and-enforce basics. It judges each message on its own rather than keeping a running strike count per person across repeated violations — if you need "three strikes and you're out" behavior, that's a worthwhile addition on top of this template.

Common questions

• Should it start in observe-only mode, warn-and-delete, or full auto-enforcement (mutes and bans)?
• What exactly should it be moderating: spam links, scams, abuse, adult content, off-topic posts, or your own rules?
• Should admins and moderators be exempt from its decisions?

Keep in mind

• Don't expect enforcement to work until the bot is an admin with delete and ban/restrict permissions granted — until then it can only warn.
• Don't switch straight to full auto-enforcement on a live, active group — run it in observe-only mode first and check its calls before trusting it with mutes and bans.
• If you need repeat offenders to face escalating consequences, know that this starter doesn't track a running strike count per person yet — that's something to add before relying on it for that.