cradlewise-bridge

An unofficial interoperability library for Cradlewise smart cribs. Uses your own Cradlewise credentials to subscribe to the live video and audio feed your crib already publishes to Cradlewise's cloud, so you can integrate it into your own monitoring, home automation, or analytics pipeline.

Status: works, alpha, single maintainer. Not affiliated with Cradlewise Inc.

What this does

• REST state polling. Read the full device shadow (temperature, baby presence, sleep state, bouncing, music, light, etc.) for each crib on your account. Builds on pycradlewise for authentication.
• Live WebRTC video + audio. Connect to Cradlewise's Janus gateway, authenticate with HMAC-signed WebSocket headers, and subscribe to your crib's H.264 video + Opus audio streams via aiortc.
• Continuous recording. Decimate video to a configurable FPS (default 2), write 5-minute MP4 segments, index them in SQLite, and purge after 48h (configurable).
• Adaptive audio spike detection. Exponential-moving-average ambient baseline + multiplier threshold + deferred post-spike capture — saves 10-second WAV snippets centered on crying/fussing/loud events.
• Multi-crib. Connect to every crib on your account concurrently with independent reconnect loops.
• Read-only by design. No control commands ever leave this library. Bouncing, music, lighting, 2-way audio — all out of scope.

Install

pip install -e .
# or
pip install -r requirements.txt

Requires Python 3.10+. Depends on pycradlewise, aiortc, aiohttp, boto3, numpy, Pillow, imageio[ffmpeg].

Quickstart

import asyncio, os
from cradlewise_bridge import CradlewiseBridge, CribConfig
from pycradlewise import CradlewiseAuth, CradlewiseClient, get_app_config

async def main():
    email, password = os.environ["CRADLEWISE_EMAIL"], os.environ["CRADLEWISE_PASSWORD"]

    # One-time discovery to map cradle_id -> device_id
    app = await get_app_config()
    auth = CradlewiseAuth(email, password, app_config=app)
    await auth.authenticate()
    cradles = await CradlewiseClient(auth).discover_cradles()

    bridge = CradlewiseBridge(
        email=email,
        password=password,
        cribs=[
            CribConfig(name=(c.baby_name or cid[:8]).lower(),
                       cradle_id=cid, device_id=c.device_id)
            for cid, c in cradles.items()
        ],
    )

    async def on_spike(crib, wav):  print(f"[{crib}] cry: {wav}")
    async def on_segment(crib, mp4): print(f"[{crib}] saved: {mp4}")
    bridge.on_audio_spike = on_spike
    bridge.on_segment_flushed = on_segment

    await bridge.run()

asyncio.run(main())

See examples/ for four runnable scripts:

Script	What it does
state_polling.py	Print the device-shadow state (no video)
connect_and_snapshot.py	Connect, grab 10 frames, save one as JPG
detect_audio_spikes.py	Save WAVs whenever crying / loud sounds occur
record_continuously.py	Full 48h rolling recording with auto-reconnect

Design notes

• The only piece that required real reverse engineering is the Janus gateway layer — WebSocket HMAC signing, videoroom handshake, trickle ICE injection into the SDP. It's documented in full in PROTOCOL.md.
• Everything above that layer (segment recording, adaptive cry detection, multi-crib supervision) is standard engineering with no Cradlewise-specific surprises. You can freely ignore those modules and just use janus.py + rest.py if you have your own media pipeline.
• pycradlewise does all the Cognito / SigV4 auth work; we don't reimplement it. Credit where due.

How is this different from pycradlewise?

pycradlewise is a REST-and-MQTT client — it reads the device shadow and exposes typed properties. It does not give you access to the video or audio stream. cradlewise-bridge adds:

• The Janus video-room activation call (AWS SigV4 REST endpoint).
• The Janus WebSocket client with HMAC-signed upgrade headers.
• The videoroom subscribe flow and trickle ICE handling.
• WebRTC peer connection via aiortc to actually receive the media.
• The recording and audio-analysis pipeline on top.

It depends on pycradlewise for authentication and shadow access.

Legal & ethical framing

This project exists so owners of Cradlewise cribs can monitor their own devices using their own credentials. It is not affiliated with, endorsed by, or supported by Cradlewise Inc.

What this code does and doesn't do

• Uses your own account. The library authenticates as you, using credentials you provide. It does not bypass authentication, impersonate other users, or access anyone else's cribs.
• Read-only by design. It subscribes to the video/audio stream and reads REST state. It never sends control commands (bouncing, music, lights, 2-way audio). The documented API exists — this project just consumes it from a non-mobile client.
• No credentials, tokens, or device identifiers are included. The signing scheme and protocol framing are documented as algorithms; the runtime config (API base, region, room secrets) is fetched from Cradlewise's own endpoints using your credentials, exactly as the official mobile app does.

Legal basis (U.S., not legal advice)

• Reverse engineering a product you own for interoperability is expressly permitted under 17 U.S.C. §1201(f) (DMCA interoperability exception) and has been broadly upheld for personal and research use (Sega v. Accolade, Sony v. Connectix).
• The Computer Fraud and Abuse Act (CFAA) targets unauthorized access. Authenticating to an account you own and accessing your own devices is authorized access by definition.
• Publishing protocol documentation derived from clean-room observation of your own authenticated sessions is protected speech; no trade secrets are misappropriated (public network traffic from a product you own is not a trade secret).

Terms of Service caveat

Cradlewise's Terms of Service may prohibit use of unofficial clients. Using this library could result in account termination at Cradlewise's discretion. That's a risk you accept by using it. If you don't want that risk, don't use this library.

Trademarks

"Cradlewise" is a trademark of Cradlewise Inc. It is used here nominatively — only to identify the product this library interoperates with — and implies no endorsement.

No warranty, not for safety-critical use

This is an independent personal-use tool provided as-is under the MIT license. Do not rely on it for safety-critical monitoring of an infant. Use the official Cradlewise app and follow their safety guidance. The authors assume no responsibility for any harm, financial loss, or account termination arising from use of this software.

Contributing

Issues and PRs welcome. Scope I'll accept:

• Bug fixes in the Janus / WebRTC flow.
• Better reconnect logic and error taxonomy.
• Additional examples (HomeAssistant bridge, Frigate sidecar, etc.).
• Protocol documentation improvements.

Scope I'll decline:

• Anything that sends control commands to a crib.
• Anything that ships credentials, tokens, or per-account identifiers.
• Anything that makes the library usable as a mass-scraping tool.

Credits

• pycradlewise by Jon Lamendola — AWS Cognito SRP auth, AWS IoT MQTT, device shadow access.
• aiortc by Jeremy Lainé — the WebRTC stack that makes Python WebRTC actually work.
• Janus Gateway — the server side of the media pipeline (not affiliated with this project, just used by Cradlewise).

Support

If this project helped you get more out of your Cradlewise, consider buying me a coffee!

License

MIT. See LICENSE.