catgenie-api

A reverse-engineered async Python library for the CatGenie AI smart litter box by PetNovations. Provides full authentication, device status, and cleaning cycle control — designed as the foundation for a Home Assistant integration.

Features

• Phone + SMS authentication — full login flow using the same OTP mechanism as the official app
• Automatic token refresh — access tokens (~30 min) are transparently refreshed using the long-lived refresh token (~10 years)
• Device status & control — start/stop cleaning cycles, read configuration, operation state, and sani-solution level
• Fully typed Pydantic v2 models — every API response is validated into typed Python objects; no raw dicts
• PEP 561 compliant — ships py.typed, zero mypy errors with strict settings
• Home Assistant ready — async context managers, shared session injection, and a clean model hierarchy that maps directly to HA entities

How It Works

The CatGenie app communicates with iot.petnovations.com (CloudFront-fronted AWS API Gateway). Every request requires three custom headers derived from an AES-CBC encrypted timestamp and HMAC-SHA256 signatures. The phone number itself is AES-encrypted in the auth request body.

There is one significant non-obvious requirement: PetNovations' edge silently discards requests from Python's standard SSL stack, returning 200 OK with an empty body instead of sending the SMS or returning data. This library uses curl_cffi to impersonate a Chrome Android TLS fingerprint, which is the only way to reliably trigger the SMS and receive API responses.

Installation

pip install catgenie

Or for development:

git clone https://github.com/kclif9/catgenieapi
cd catgenie-api
pip install -e ".[dev]"

Usage

Authentication

import asyncio
from catgenie import CatGenieAuth, CatGenieClient

async def main():
    auth = CatGenieAuth()
    # Step 1 — trigger SMS to the phone number
    await auth.request_login_code(country_code=61, phone="499999999")

    # Step 2 — exchange the SMS code for tokens
    code = input("Enter SMS code: ")
    credentials = await auth.login(country_code=61, phone="499999999", code=code)
    await auth.async_close()

    # Step 3 — use the client
    client = CatGenieClient(credentials)
    try:
        devices = await client.get_devices()
        for device in devices:
            print(f"{device.name}: {'online' if device.is_online else 'offline'}")
            print(f"  Sani-solution remaining: {device.remaining_sani_solution}%")
            print(f"  Lifetime cycles: {device.configuration.total_cycles}")
    finally:
        await client.async_close()

asyncio.run(main())

Controlling a Device

client = CatGenieClient(credentials)
try:
    devices = await client.get_devices()
    device = devices[0]

    # Start a cleaning cycle
    await client.start_cleaning(device.manufacturer_id)

    # Stop a cleaning cycle
    await client.stop_cleaning(device.manufacturer_id)
finally:
    await client.async_close()

Token Persistence

import json, dataclasses
from catgenie import Credentials

# Save after login
with open("credentials.json", "w") as f:
    json.dump(dataclasses.asdict(credentials), f)

# Restore on next run
with open("credentials.json") as f:
    credentials = Credentials(**json.load(f))

Shared Session (Home Assistant pattern)

from curl_cffi.requests import AsyncSession, Response
from catgenie import CatGenieAuth, CatGenieClient
from catgenie.const import TLS_IMPERSONATE

session = AsyncSession(impersonate=TLS_IMPERSONATE)
try:
    auth = CatGenieAuth(session=session)
    credentials = await auth.login(...)

    client = CatGenieClient(credentials, session=session)
    devices = await client.get_devices()
finally:
    await session.close()

API Reference

CatGenieAuth

CatGenieClient

Key Models

Project Structure

src/catgenie/
├── __init__.py     # Public API exports
├── py.typed        # PEP 561 marker
├── auth.py         # Phone + SMS authentication (CatGenieAuth, Credentials)
├── client.py       # High-level async client (CatGenieClient)
├── const.py        # Constants and all endpoint paths
├── models.py       # Pydantic v2 models for every API response
└── signing.py      # AES-CBC encryption + HMAC-SHA256 request signing

Prior Art & Credits

• Rukongai/CatDjinni — reverse engineering research: signing algorithm, Frida scripts
• appkins/hass-catgenie — earlier HA integration (requires manual refresh token)
• PrimeAutomation/petnovations — earlier HA integration (same limitation)

License

MIT