Telegram Autopilot

---

name: telegram-autopilot description: Manage a Telegram userbot autopilot that responds to private messages as the user using AI. Use when the user wants to set up auto-replies on their personal Telegram account, manage allowed contacts, configure AI response style, or send messages/media as themselves. Triggers on "telegram autopilot", "auto reply telegram", "manage my telegram", "respond for me", "telegram userbot", "paid media telegram". Requires secrets — Telegram API credentials (api_id, api_hash), phone number, optional 2FA password, AI provider API key (Anthropic or OpenAI-compatible), optional Telegram bot token for owner notifications.

---

Telegram Autopilot

> Source: https://github.com/Shor73/telegram-autopilot > Author: @Shor73 > License: MIT

AI-powered autopilot for personal Telegram accounts. Responds to private messages as the user when they're unavailable.

Prerequisites

Python 3.10+

Telethon: pip3 install telethon

Telegram API credentials (api_id + api_hash from https://my.telegram.org)

An Anthropic API key (for AI responses)

Setup

1. Get Telegram API Credentials

Direct the user to https://my.telegram.org → API Development Tools → create app. They need: API ID (number) and API Hash (string).

2. Login Flow

Telegram login requires: phone → OTP code → optional 2FA password.

Critical: OTP codes expire in ~60 seconds. Minimize latency:

Use a file-based code exchange (script polls a file every 200ms)

Or serve a simple web form on a local port for instant code entry

Never rely on chat round-trip for code delivery — too slow

Run scripts/setup.py with the user's credentials. It handles: 1. Requesting the OTP code 2. Waiting for code via file (enter_code.txt) 3. 2FA password if needed 4. Saving the session file

python3 scripts/setup.py --api-id 12345 --api-hash "abc123" --phone "+1234567890"

The session file (.session) persists auth — login is one-time only.

3. Configure Contacts

Edit config.json to define allowed contacts:

{
  "contacts": {
    "username": {
      "name": "Display Name",
      "id": 123456789,
      "tone": "friendly",
      "language": "en"
    }
  },
  "ai": {
    "provider": "anthropic",
    "model": "claude-sonnet-4-6",
    "api_key": "sk-ant-...",
    "max_tokens": 300
  },
  "owner": {
    "name": "Owner Name",
    "bio": "Brief description for the AI persona",
    "telegram_id": 123456789
  },
  "notifications": {
    "bot_token": "optional-bot-token-for-notifications",
    "chat_id": "optional-chat-id"
  }
}

4. Start Autopilot

python3 scripts/autopilot.py --config config.json --session session_name

Features

Auto-Reply

Responds only to allowed contacts in private chats

Marks messages as read before replying (natural behavior)

Simulates typing delay proportional to response length

Maintains conversation history for context

AI Persona

System prompt configurable per-contact (tone, language)

Never invents facts — says "I'll check and get back to you" for unknowns

Never reveals it's AI

Notifications

Forwards received messages to owner via Telegram bot

Forwards sent replies for monitoring

Paid Media (Channels Only)

Telegram paid media (inputMediaPaidMedia) only works in channels, not private chats.

To send paid media: 1. Create a private channel 2. Post media with InputMediaPaidMedia(stars_amount=N, extended_media=[...]) 3. Generate invite link and send to recipient

python3 scripts/send_paid_media.py --session session_name --target username --photo /path/to/photo.jpg --stars 1

Management Commands

Stop autopilot: kill the process or send SIGTERM. Add/remove contacts: edit config.json and restart.

Architecture

telegram-autopilot/
├── SKILL.md
├── scripts/
│   ├── setup.py          — Login flow (OTP + 2FA)
│   ├── autopilot.py      — Main event loop
│   ├── send_paid_media.py — Paid media via channel
│   └── code_server.py    — Web form for fast OTP entry
└── references/
    └── telegram-auth.md  — Telegram auth flow documentation

Required Credentials

| Secret | Purpose | Where used | |---|---|---| | Telegram API ID + Hash | MTProto client auth | setup.py, autopilot.py | | Phone number + 2FA | Account login (one-time) | setup.py | | AI API key | Response generation | autopilot.py | | Bot token (optional) | Owner notifications | autopilot.py |

All secrets are stored in config.json — never commit this file.

Security & Ethics

Session security: The .session file grants full access to the account. Protect it like a password.

Transparency: The AI is instructed to be honest if directly asked whether it's AI.

OTP server: code_server.py binds to 127.0.0.1 (localhost only). Never expose it to the network.

Notifications: The skill can forward messages to the owner via bot. Ensure you control the bot token and chat_id.

Rate limits: Telegram may restrict accounts with aggressive automation. The autopilot uses natural delays.

One session at a time: Only one process can use a session file. Stop autopilot before running other scripts.

Platform policies: Using a userbot to auto-reply may violate Telegram ToS. Use at your own risk.

1. Get Telegram API Credentials

Direct the user to https://my.telegram.org → API Development Tools → create app. They need: API ID (number) and API Hash (string).

2. Login Flow

Telegram login requires: phone → OTP code → optional 2FA password.

Critical: OTP codes expire in ~60 seconds. Minimize latency:

Use a file-based code exchange (script polls a file every 200ms)

Or serve a simple web form on a local port for instant code entry

Never rely on chat round-trip for code delivery — too slow

Run scripts/setup.py with the user's credentials. It handles: 1. Requesting the OTP code 2. Waiting for code via file (enter_code.txt) 3. 2FA password if needed 4. Saving the session file

python3 scripts/setup.py --api-id 12345 --api-hash "abc123" --phone "+1234567890"

The session file (.session) persists auth — login is one-time only.

3. Configure Contacts

Edit config.json to define allowed contacts:

{
  "contacts": {
    "username": {
      "name": "Display Name",
      "id": 123456789,
      "tone": "friendly",
      "language": "en"
    }
  },
  "ai": {
    "provider": "anthropic",
    "model": "claude-sonnet-4-6",
    "api_key": "sk-ant-...",
    "max_tokens": 300
  },
  "owner": {
    "name": "Owner Name",
    "bio": "Brief description for the AI persona",
    "telegram_id": 123456789
  },
  "notifications": {
    "bot_token": "optional-bot-token-for-notifications",
    "chat_id": "optional-chat-id"
  }
}

4. Start Autopilot

python3 scripts/autopilot.py --config config.json --session session_name