omnichat 1.0.0

Direct omnichannel chat SDK for Python — Telegram, Slack, Google Chat, Teams, WhatsApp. No server needed — install, connect your channels, handle messages.

OmniChat

Omnichannel chat SDK for Python. Connect Telegram, Slack, Google Chat, Microsoft Teams, and WhatsApp with a single package — no server deployment needed.

Install

pip install omnichat

Usage

import asyncio
from omnichat import OmniChatBot

async def main():
    sdk = OmniChatBot()
    
    sdk.use_telegram("your-bot-token")
    # sdk.use_slack("xoxb-bot-token", "xapp-app-token")
    # sdk.use_gchat(service_account_key="/path/to/service-account.json")
    # sdk.use_teams("your-app-id", "your-app-password")
    
    async def on_message(thread, msg):
        print(f"[{msg.platform}] {msg.author.user_name}: {msg.text}")
        await thread.reply(f"You said: {msg.text}")
    
    async def on_mention(thread, msg):
        await thread.reply(f"Hey {msg.author.user_name}! You mentioned me.")
    
    sdk.on_message(on_message)
    sdk.on_mention(on_mention)
    
    await sdk.start()

asyncio.run(main())

Live-updating replies (tracked messages)

For streaming AI output, don't wait for the full response before showing anything — send once, then rewrite that same message in place as more text arrives:

handle = await thread.reply_tracked("Thinking")  # appears immediately
await thread.update_reply(handle, "Thinking about it...")  # rewrites the same message
await thread.update_reply(handle, "Here's the full answer.")

This avoids a chat platform showing "not responding" during long generations, and avoids leaving a stray "Thinking..." message behind — there's only ever one message, and it's rewritten as it grows.

Support varies by platform:

Platform	reply_tracked()	update_reply()
Google Chat	✅	✅ — PATCH messages.update
Slack	✅	✅ — chat.update
Telegram	✅	✅ — editMessageText
Microsoft Teams	✅	✅ — Bot Framework activity update
WhatsApp	✅ (returns a real message id)	❌ raises NotImplementedError — the Cloud API has no endpoint to edit a sent message

Supported Platforms

Platform	Method	How it works
Telegram	use_telegram(bot_token)	Long polling — no public URL needed
Slack	use_slack(bot_token, app_token)	Socket Mode — no public URL needed
Google Chat	use_gchat(service_account_key, port, path, upload_as_user)	HTTP webhook + service-account app auth
Microsoft Teams	use_teams(app_id, app_password)	HTTP webhook + Azure OAuth2
WhatsApp	use_whatsapp(access_token, phone_number_id, verify_token, port, path)	HTTP webhook + Graph API

Credentials

Platform	What you need	Where to get it
Telegram	Bot Token	Talk to @BotFather on Telegram
Slack	Bot Token + App Token	api.slack.com — enable Socket Mode
Google Chat	Service account JSON key (+ HTTPS endpoint registered in the Chat API config)	console.cloud.google.com — enable the Chat API, configure the app, create a service account
Microsoft Teams	App ID + App Password	portal.azure.com — Register a Bot
WhatsApp	Access Token + Phone Number ID + Verify Token	developers.facebook.com — create a Meta Business App with WhatsApp integration

Google Chat file attachments (send_file()) additionally need upload_as_user set to a Workspace user your service account has domain-wide delegation for — the Chat API's attachment upload endpoint only accepts user authentication, not the app (service-account) authentication used for text messages.

Requirements

• Python >= 3.10
• No server deployment needed
• No database needed

Testing

The SDK includes unit tests for core logic (history, state, fluent chaining, etc.):

pytest tests/

Note: Network-level tests for real platform integrations are not included. To verify platform integration, deploy the package and test against real platform APIs (Telegram Bot API, Slack Web API, Google Chat API, etc.) with valid credentials and a public URL tunnel (ngrok, Cloudflare tunnel, etc.).