Hello everyone!
We’ve passed the 100 mark! 12 days ago, I challenged you to 100 integrations. This morning, the store index has 102, not even counting Frigate by @lmilcent, which was released after the indexing.
| September 27 | Today | |
|---|---|---|
| Integrations in the store | 91 | 102 |
| Contributors | 20 | 22 |
| Integrations ≥ 5% in HA covered | 34/48 | 36/48 |
In 12 days: HomeKit Controller (@prohand, the biggest coverage lever left), Forecast.Solar, Broadlink, Mammotion, and TV Guide (@prohand again!), Jellyfin & Emby and Android TV Remote (@guim31), Frigate (@lmilcent), Pushover (@cicoub13), Tydom (requested since July), Orbit B-hyve, and Daikin Airbase. Welcome to the two new contributors, quentinlegay and plsoftware
and thank you all!
Next goal: North America 

By redoing the analysis, I looked at a figure I had never looked at before: out of 696,234 active Home Assistant installations, 17% are in the United States (118,589). It’s the 2nd country after Germany (19%), and more than 3 times France (5.3%). With Canada (2.4%), North America represents almost 1 in 5 installations.
On the store side, we’re very strong on what no global platform will ever do: GRDF, VigiEau, Fuel Prices, AlloCiné, TCL… and we must continue! But an American who discovers Gladys today looks for their Ring doorbell, Lutron switches, Nest thermostat, and Roku… and doesn’t find them. We give the image of a « 100% French-French » project, whereas Gladys is translated into English and the store already requires English documentation for each integration.
We’re not starting from scratch: SmartThings, HomeKit Controller, Apple TV, Samsung/LG/Android TVs, Plex, Sonos, Kasa, Hue, Enphase, SolarEdge, EcoFlow, Hydro-Québec… and our first 100% American brand, Orbit B-hyve
What’s missing are the brands found in every American home.
« But I don’t have the hardware! » → develop blindly
This is what I’m proposing again: develop an integration without having the device. With a documented API, an existing library, and Claude, it’s quite feasible. No need to wait for a tester to publish: the day someone with the device tries it, they will come back to you with their feedback. In the meantime, it fills the store and shows that we are open to American users.
I’ve classified the candidates into three levels:
zero hardware: public API or free account, 100% testable from France;
hardware sold in France: someone here might have one;
blind: we code from France, feedback will come from those who have the device.
The % = share of Home Assistant installations using the integration (analytics from October 9), or number of HACS installations. HA does not publish these stats by country: for a service that only exists in North America, I give an order of magnitude by relating the % to the country’s weight.
Zero hardware (perfect for a first integration)
| Integration | HA Usage | Type/access | Why |
|---|---|---|---|
| NWS (National Weather Service) | 1.9%, ≈ 1 US install out of 9 | weather, free API without key | The American Météo France: forecasts + official alerts (tornadoes, hurricanes, heatwaves) as scene triggers |
| Environment Canada | 0.6%, ≈ 1 Canadian install out of 4 | weather, free and bilingual API | The Canadian counterpart, alongside Hydro-Québec |
| Pirate Weather | 8,500 HACS | weather, free key | The successor to Dark Sky, worldwide |
| Team Tracker (ESPN scores) | 5,200 HACS | provider, public API | « Current match » widget + « my team scores » trigger: NFL, NBA, MLB, NHL… and Ligue 1 |
| Twilio | 0.3% | communication, free trial account | SMS and calls, the US standard for notifications |
| Slack / Discord | 0.1% / 0.5% | communication | Testable by anyone |
Hardware sold in France (you might already have it)
| Integration | HA Usage | Access | Note |
|---|---|---|---|
| Ring | 6.3% | Cloud, Node lib ring-client-api (the one from homebridge-ring) |
The biggest missing « US » item: doorbell as scene trigger, camera images |
| Alexa (control Echo devices) | 5% + 29,900 HACS | Unofficial cloud, Node lib alexa-remote2 |
The opposite of the current Alexa integration: make Gladys speak on Echo devices (announcements, volume, playback) |
| Google Nest | 3.8% | Official SDM API (one-time $5) | Thermostats, cameras, doorbells |
| Govee | 2.7% LAN + 2.5% BLE + 11,100 HACS | Official LAN API (UDP) | Huge in the US, quick win |
| Eufy Security | 9,700 HACS | Node lib eufy-security-client |
Cameras and doorbells without subscription |
| Blink | 1.7% | Unofficial cloud | Amazon cameras |
| Tesla (car + Powerwall) | 0.9% + 0.4% + 4,400 HACS | Fleet API or Tessie | Fits Gladys’ energy axis |
Blind (we code here, feedback will come from those who have the device)
| Integration | HA Usage | Access | Note |
|---|---|---|---|
| Lutron Caséta | 2.4% | Local (LEAP), Node lib lutron-leap |
The US lighting reference, the biggest North American gap |
| Roku | 2.2% | Local, ECP API documented by Roku | US streaming #1, completes Samsung/LG/Android TV/Apple TV |
| Wemo | 1.8% | Local (UPnP) | Belkin cut its cloud on January 31, 2026: its users are looking for a local solution |
| Rachio | 1.3% | Official API (key in the app) | Irrigation, alongside B-hyve |
| August / Schlage | 1.1% / 0.6% | Cloud | Locks |
| Bond | 0.9% | Local, documented API | Ceiling fans (essential in the US) and RF shades |
| Vizio | 0.7% | Local | TV |
| Opower | 0.7% | Cloud | Consumption of dozens of US providers (PG&E, ConEd…): the American Enedis, directly in energy tracking |
| Honeywell (Total Connect Comfort) | 0.6% | Cloud | Thermostats |
| Sense / Emporia Vue | 0.4% / 4,800 HACS | Cloud | Electrical panel energy monitors |
| Sleep Number | 0.4% | Cloud | Connected beds: presence in bed as a trigger |
| Rheem EcoNet | 0.35% | Cloud | Water heaters: the « water heater » category already exists in Gladys |
And on HACS: Wyze (3,500), GE Appliances (3,700), Hubspace, the Home Depot brand (2,200).
ecobee (1.6%): the official API has been closed to new developers since March 2024… but ecobee thermostats speak HomeKit, so @prohand’s HomeKit Controller should already cover them locally. We’re looking for testers to confirm this!
The « blind » method
- Announce yourself here: « I’m taking Roku (blind) », to avoid duplicates.
- Start with what exists: the official API documentation, the Home Assistant integration (Apache 2.0 license), and very often, a Homebridge plugin or an existing npm library in Node. Give all that to Claude, he carries it very well. Respect licenses and credit authors.
- Simulate the device: HA tests often contain real device responses (« fixtures » folders), and GitHub issues are full of « diagnostics » files. Make mocks out of them, and write tests that run the integration end-to-end without the device.
- Think American: switch your Gladys profile to °F and miles for testing, take care of the
docs/en.md(this is what an American will read first), and put an English summary at the top of your Development thread. - Publish in beta, honestly: write « developed without the hardware, feedback welcome » in the documentation. The day a user with the device tests it, they will come back to you via GitHub issues, and you will correct based on their logs.
Still open
The rest of the top HA is still looking for takers: DLNA DMR / DMS (27.9 % / 20.4 %), Brother (8.7 %), WLED (6.1 %), BTHome (5.4 %), ONVIF (5.2 %), and on the HACS side, Sonoff / eWeLink locally (number 1 with 45,000 installations). And same logic as for the US: Germany is number 1 for HA (19 %), and AVM FRITZ!Box accounts for 8.8 % + 6.5 % ![]()
Goal: 10 North American integrations by the end of November
Take one, even a small one (NWS or Team Tracker, it’s an afternoon with Claude), reply « I’ll take X » here, then open a thread in Development when it’s ready to test.
For the rest, nothing changes: official repo template (« Use this template ») + Node SDK, everything is in the developer documentation. Publishing = tagging your public repo with the topic gladys-assistant-integration, no approval necessary. Prioritize local over cloud.
Don’t develop? You have a Ring, an Echo, a Nest, Govee, or an ecobee? Say so here: you’ll be the first testers ![]()