Telegram Bot Overview

The Telegram bot is an optional component that runs as a daemon thread inside the Studio backend process. It uses the python-telegram-bot library with manual polling (not webhooks) and communicates with the Studio backend via localhost HTTP calls.

Architecture

The bot runs in a background thread with its own asyncio event loop. This is necessary because the main thread runs uvicorn's async event loop, and python-telegram-bot's run_polling() requires the main thread. The solution:

1. A daemon thread is created with threading.Thread(daemon=True)
2. The thread creates a new asyncio.event_loop
3. The bot application is built and handlers are registered
4. Manual polling starts: initialize() -> start() -> updater.start_polling()
5. The polling loop runs until _bot_started_at is set to None (stop signal)

Communication with Studio

The bot does not import or call Studio backend functions directly. Instead, it makes HTTP requests to http://127.0.0.1:{STUDIO_PORT} with an X-API-Key header, using the same API that external clients use. This keeps the bot fully decoupled from the backend internals.

API calls used by the bot:

Endpoint	Purpose
GET /api/admin/presets	List available presets
GET /api/admin/presets/{id}	Get preset details
GET /api/admin/presets/{id}/placeholders	Extract placeholder questions
POST /api/admin/presets/{id}/run	Execute a preset
GET /api/run/status/{prompt_id}	Poll job progress
GET /api/comfyui/view	Download the result file

Auto-start

On backend startup, telegram_bot.auto_start() is called. If TELEGRAM_BOT_TOKEN is set in the environment (or in the database settings), the bot starts automatically. If the token is not set, the bot remains stopped and can be started later from the Settings page.

Dependencies

The bot requires the python-telegram-bot package. If the package is not installed, the bot control endpoints return an error message instead of crashing the backend.