🚀 Telegram Web MCP Server

A powerful, token-efficient Model Context Protocol (MCP) server that lets AI agents interact with Telegram Web — without wasting tokens on DOM trees and screenshots.

🤔 The Problem

Existing approaches to Telegram automation for AI agents either:

• Require Telegram Bot API tokens (limiting you to bot-only features)

• Use heavy browser automation that wastes thousands of tokens parsing HTML DOMs and analyzing screenshots

• Provide no persistent session, requiring re-authentication every time

Related MCP server: telegram-mcp

✅ The Solution

This MCP server provides a clean, text-based API layer over a persistent Playwright browser session. Your AI agent gets structured text responses instead of raw HTML — cutting token usage by 10-50x compared to generic browser automation.

✨ Features

Key Design Decisions

• 🔋 Token Efficient — Returns clean text, not complex HTML DOMs

• 💾 Persistent Session — Log in once, session is saved permanently

• 🤖 AI-Ready — Includes dedicated skill guides for Claude, Cursor, and Antigravity

• 🔒 Local-First — Runs on your machine, no data sent to third parties

📦 Installation

git clone https://github.com/Fayzullo637/mytelegrammini1-mcp.git
cd mytelegrammini1-mcp
npm install
npx playwright install chromium

🚀 Usage

With Claude Desktop / Antigravity

Add to your MCP config:

{
  "mcpServers": {
    "telegram": {
      "command": "npx",
      "args": ["tsx", "src/index.ts"],
      "cwd": "/path/to/mytelegrammini1-mcp"
    }
  }
}

Then ask your AI: "Search for @username on Telegram and send them a message"

🛠️ Tech Stack

• TypeScript — Type-safe server implementation

• Playwright — Reliable browser automation with persistent contexts

• MCP SDK — Standard Model Context Protocol interface

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📄 License

This project is licensed under the MIT License — see the LICENSE file for details.