LyAdmin: Anti-Spam for Your Group

You open your group and find ads for casinos and easy work from fresh accounts. Cleaning it up by hand gets old within a week. LyAdmin takes over that chore: the bot watches messages and removes spam while you get on with real conversation. 🧹

Who built it and why that matters

Open code also means a long history of public fixes you can read.

The bot was written by developer Yuri Lyubchak (known as LyoSU), and the source code is public on GitHub. That is unusual among anti-spam bots: the rules are visible, and you can audit them or run your own copy.

The current version is a ground-up rewrite in TypeScript that talks over the MTProto protocol directly instead of the regular Bot API.

How it decides something is spam

Under the hood is a pipeline. First it gathers signals about the message and its author. From them it computes a spam probability score, and a policy decides what to do: keep, delete or ask admins for help. 🔬

The bot can use extra services to analyze meaning if the owner connects them. Without them, those stages are simply skipped.

Reports and quiet mode

• Trusted reporter: if such a member reports a stranger's message, it is removed, the author is muted for an hour, and the vote still opens.
• Quiet mode: off by default. When enabled, the bot speaks up only when it is unsure or an admin is needed.

That keeps your group from filling with service messages.

Five commands are enough

• /start: begin.
• /help: command reference.
• /stats: how much spam was caught.
• /mystats: your personal numbers.
• /lang: interface language.

Setup in three steps

Connect it during the day while you are online to check the first actions.

1. Add the bot to your group.
2. Give it admin rights to delete messages and restrict members.
3. Use /help to explore the settings available in a private chat with the bot.

Without admin rights the bot can see spam but cannot remove it. 🔧

Things to plan for

Tell fellow admins before switching it on so nobody is surprised by a muted member.

Every anti-spam tool gets it wrong sometimes. A new member with an odd first message can look suspicious. Watch the stats for the first week and soften the settings if needed.

Mind privacy too: the bot reads group messages, so tell members in a pinned post. The public source code lets you check how data is handled.