Skip to content
Back to skills

Telegram Bot

ASecurity

Creates a Telegram bot project from a short description — handlers, menu buttons, storage, admin commands, optional AI-model replies — with the token in .env, tests, and instructions to run it 24/7. Use when the user wants a Telegram bot or says "сделай бота", "телеграм-бот", "бот для заявок", "telegram bot".

  • 2 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 13, 2026
ai-agentssqldockergitapi

Works with

  • api

Security analysis

A100/100

Scanned October 13, 2026

npx -y skills add dmminga1983/minga --skill telegram-bot --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Telegram Bot?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Telegram Bot
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/dmminga1983-telegram-bot/badge)](https://www.skillsdirectory.com/skills/dmminga1983-telegram-bot)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: telegram-bot
description: Creates a Telegram bot project from a short description — handlers, menu buttons, storage, admin commands, optional AI-model replies — with the token in .env, tests, and instructions to run it 24/7. Use when the user wants a Telegram bot or says "сделай бота", "телеграм-бот", "бот для заявок", "telegram bot".
argument-hint: "[what the bot should do]"
---

# Telegram bot / Телеграм-бот

Idea: $ARGUMENTS

Reply in the user's language; bot texts in its users' language.

## 1. Spec (one screen)

Purpose in one sentence · users · commands and buttons (`/start`, `/help`, main menu) · data stored · notifications to the admin · external APIs or AI model · where it will run.

## 2. Build

Delegate to the `bot-builder` agent with the spec. Required in every bot:
- `.env.example` with `TELEGRAM_BOT_TOKEN=` and `ADMIN_IDS=`; `.env` in `.gitignore` — run `/minga:secrets-setup`;
- startup check for missing variables, no token in logs;
- handlers for `/start`, `/help`, unknown input and errors;
- input validation and rate limiting for expensive actions;
- if an AI model is used: per-user daily limit and max tokens per reply;
- SQLite storage by default; minimum personal data;
- tests for the handler logic (without calling Telegram).

## 3. Run

Instructions in the README: create the bot in **@BotFather** → copy the token into `.env` → install → run. For 24/7: a systemd unit or a `Dockerfile` + `docker compose`, with restart on failure.

## 4. Before publishing

`/minga:secure-publish`. A leaked token lets anyone control the bot — if it ever leaks, revoke it in @BotFather (`/revoke`).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.