Telegram Feedback Bots Builder

A self-hosted toolkit for provisioning and operating Telegram feedback bots. The new stack pairs a Django + python-telegram-bot backend with a SvelteKit mini app so owners can manage bots, rotate tokens, review stats, and handle banned users without leaving Telegram. Think of it as Livegram but more powerful.

Features

Builder Bot

• /start launches the inline menu with a deep link to the mini app.
• /broadcast forwards announcements to builder users with optional filters (joined before/after a date, active in the last N days, username-only, sample every Nth user).
• /whitelist, /manage, /restart, and /update remain available for administrators.

Mini App Dashboard

• Create feedback bots after validating their tokens against BotFather.
• Update start / receipt messages and link or unlink groups.
• Rotate bot tokens securely with enforced admin approval before re-enabling (when configured).
• Inspect live stats, review audit history of broadcasts, and manage banned users per bot.
• Decide which media types (photos, videos, voice, files, stickers) each bot accepts before forwarding.
• Choose a communication mode per bot (standard, private, or anonymous) to control how user details appear to moderators and in forwarded threads.
• Toggle and tune anti-flood protection to cap incoming messages at a configurable interval and warn users when throttled.
• Manage builder access: whitelist, language, admin flags, and removal.

🔒 See the Privacy Policy for details on data that is stored.

Feedback Bots

• Route every user conversation into its own forum topic (or the owner chat) for clean threading.
• Forward edits, mirror replies back to the user, and automatically acknowledge receipt with an emoji.
• Politely decline disallowed media with a localized reply while leaving text enabled.
• Provide owner tooling: /broadcast, /ban, /unban, /banned, /delete (reply to remove a single pair), and /clear (reply to drop that message and everything after it in the thread), plus /start for users. Broadcasts can reuse the same filters as the builder bot.

Broadcast Segmentation Cheat Sheet

Reply to the message you want to send with /broadcast and put filters on the following lines of the same command message. Leave the command by itself (or add done) to reach everyone:

joined_after 2024-01-01
joined_before 2024-05-01
active_within 14
username_only yes
sample_every 3

Dates accept YYYY-MM-DD or ISO datetimes. active_within is expressed in days and maps to recent feedback activity. sample_every keeps every Nth chat after other filters run.

Technology Stack

• Backend: Python 3, Django 5, Ninja API, python-telegram-bot 21, Granian (ASGI), SQLite (Postgres optional).
• Frontend: SvelteKit, Vite, Tailwind CSS, Bits UI, Telegram Mini App SDK.
• Packaging & Tooling: mise, uv, pnpm, Docker, lefthook.
• Infrastructure extras: Traefik override (optional), encrypted token storage (Fernet).

Setup

1. Clone the repository

   git clone https://github.com/yshalsager/telegram_feedback_bot.git
   cd telegram_feedback_bot

2. Install toolchain with mise

   mise install

3. Create .env (values mirror mise.toml[env]). At minimum:

   DJANGO_SECRET_KEY="$(mise r generate-secret-key)"
   TELEGRAM_BUILDER_BOT_TOKEN="123456789:AA..."
   TELEGRAM_BUILDER_BOT_WEBHOOK_URL="https://your-domain.com"
   TELEGRAM_BUILDER_BOT_WEBHOOK_PATH="telegram"
   TELEGRAM_BUILDER_BOT_ADMINS="123456789,987654321"
   TELEGRAM_ADMIN_CHAT_ID="-10000000000"
   TELEGRAM_LOG_TOPIC_ID=1
   TELEGRAM_ENCRYPTION_KEY="$(mise r generate-encryption-key)"
   TELEGRAM_LANGUAGES="en ar"
   POSTGRES_DB="feedback_bot"
   POSTGRES_USER="feedback_bot_user"
   POSTGRES_PASSWORD="super-secret"
   DJANGO_ALLOWED_HOSTS="localhost 127.0.0.1"
   TELEGRAM_NEW_BOT_ADMIN_APPROVAL=true

   Set TELEGRAM_BUILDER_BOT_WEBHOOK_URL to the public HTTPS origin serving Granian. Leave DJANGO_USE_SQLITE=true if you prefer SQLite for local experiments.

4. Install dependencies

   mise x uv -- uv sync
   mise x pnpm -- pnpm install

5. Run migrations

   mise x uv -- uv run manage.py migrate

Running

Docker

The compose stack bundles the backend, PTB workers, and Postgres. Copy your .env next to docker-compose.yml, then:

docker compose up --build -d

Static assets from the SvelteKit build are served by Granian under /app/. Attach volume data/ if you need to persist exported artifacts.

Manual (mise + uv + pnpm)

Run backend and frontend separately during development:

# Backend API + Telegram bots
mise x uv -- uv run granian --interface asgi config.asgi:application

# Frontend mini app
mise x pnpm -- pnpm run dev

Or let mise orchestrate everything:

mise run migrate
mise run api-dev   # starts Granian with autoreload
mise run client-dev

For a production build of the mini app:

mise x pnpm -- pnpm run build

The compiled bundle lands in build/ and is mounted automatically by Granian when GRANIAN_STATIC_PATH_* variables are set (default values already point there).

Telegram Webhook Notes

• Set the builder webhook by exposing Granian at https://<domain>/api/webhook/<TELEGRAM_BUILDER_BOT_WEBHOOK_PATH>/.
• Each managed bot receives a dedicated webhook at /api/webhook/<bot_uuid>/; secrets are generated from TELEGRAM_ENCRYPTION_KEY.
• When admin approval is enabled (TELEGRAM_NEW_BOT_ADMIN_APPROVAL=true), new bots start disabled until a builder admin activates them from the dashboard.

Testing

# Backend tests
mise x uv -- uv run pytest

# Frontend unit tests
mise x pnpm -- pnpm run test

# Linting & type checks
mise x pnpm -- pnpm run lint
mise x pnpm -- pnpm run check

Migrating from the legacy Pyrogram version

1. Create backups of your existing feedback_bot.db builder database and the per-bot files inside data/ (or whichever paths you configured).
2. Populate environment variables used by the legacy stack. Export the old ENCRYPTION_KEY (the XOR key from the Pyrogram build) alongside the new .env variables.
3. Install and migrate the Django project as described above, then run:

   mise x uv -- uv run manage.py migrate_legacy_data --builder-db feedback_bot.db --data-dir data

   Adjust the paths if your legacy assets live elsewhere. The command can be re-run safely; it upserts records.
4. Review migrated bots in the mini app. Re-enable each bot once you are satisfied (admin approval still applies when TELEGRAM_NEW_BOT_ADMIN_APPROVAL=true).
5. Reset webhooks by redeploying the app or restarting granian so the builder and managed bots pick up their new webhook endpoints.

Project Layout

• config/ – Django project configuration and ASGI entrypoint.
• feedback_bot/ – Backend app (API, models, telegram builders, PTB handlers).
• src/ – SvelteKit mini app served inside Telegram.
• data/ – Shared volume for runtime assets (bot storage, exports).
• docker-compose.yml – Production-ready stack with Postgres + Granian.
• mise.toml – Task runner and environment defaults.

Historical Versions

• Original Pyrogram-based builder
• Standalone single-bot edition

Acknowledgements

• Django, Ninja, and Granian for the backend.
• python-telegram-bot for Telegram integrations.
• SvelteKit, Tailwind CSS, and Bits UI for the mini app.
• uv and mise for fast, reproducible environments.

Development Tips

• mise r starts migrations, Granian, and the SvelteKit dev server in one go.
• Use mise x uv -- uv run manage.py shell_plus for interactive debugging.
• Internationalization helpers live under src/locales and feedback_bot/locale; run mise r i18n_update after editing strings.