# Internet-of-Trash: a list of smart home devices to avoid buying

> Internet-of-Trash is a short, opinionated list of consumer IoT products that have broken, restricted or never supported local control, and the reason it is worth reading is the pattern in its entries: a cloud service ends, a vendor adds a firmware authorisation change, and a device that worked with Home Assistant stops working with nothing in the code changing.

**unixorn/internet-of-trash** — A list of IOT devices that are garbage

- Repository: https://github.com/unixorn/internet-of-trash
- Stars: 559 · Forks: 13
- Language: Unknown
- License: NOASSERTION
- Published: 2026-10-22 · Updated: 2026-10-22 · Language: en
- Canonical page: https://hysenlabs.com/projects/unixorn-internet-of-trash

## What the list is for, and who writes it

Internet-of-Trash is a list of IOT devices that are garbage, and the README is blunt in a way that is unusual for a project: a list of IOT devices to avoid.

The author frames it in terms of Home Assistant. The topics list home-assistant, home-automation, hubitat, hass, iot and iot-device, and the main table is titled the Hall of Shame. The entries are all products that a Home Assistant user might otherwise buy.

So the value is narrow and specific rather than general. If you are not running a local automation system, a list of which smart bulbs have a driver whine is of no interest to you. If you are, it is a shortlist of devices that will cost you time.

The authorship matters too, because every entry is a first-person account. The Amysen RGBW bulbs entry describes buying about ten of them over three months and finding only three still in service. The Aqara Vibration Sensor entry describes a pairing procedure involving holding a button until the lights flash, then pressing it repeatedly every second or two, and a sensor that kept falling off the Zigbee mesh. These are reports from someone who installed the thing, not a specification sheet.

The primary language is recorded as unknown, the licence field is a custom one GitHub cannot classify, the default branch is main, and the last push was on 2026-08-08 with the project not archived. There are no GitHub releases, which is consistent with a repository whose content is a table rather than a program.

The repository itself is minimal: .github/, .gitignore, .pre-commit-config.yaml, LICENSE and README.md. There is no code, no tooling beyond a pre-commit configuration, and no data file separate from the document. The list is the document.

## The three criteria for getting on the list

The list of reasons is short and specific, and it is worth reading because the criteria are about vendor behaviour rather than about hardware quality.

A device had an open API, then switched to requiring you to use their application instead of directly scraping the API. That is the first criterion, and it is the most consequential one. The device worked, and then the vendor changed the terms of access so that the only supported path was their own app.

Devices which are cloud-only is the second. If the product has no local control path at all, then it cannot be integrated with anything, and the failure is complete rather than partial.

The third is the most demanding: anything that requires you to use their application and does not document their API so it can be integrated with other ecosystems. So a cloud-only device that also documents nothing gets listed, and the combination of a closed API and no documentation is the specific failure the list exists to flag.

Those three criteria share a theme. None of them is about the device being cheap, badly made, or full of security holes. They are all about what happens to your control over hardware you have already bought. A device that stops working because of a server shutdown is not defective, it is orphaned, and that is what the list is tracking.

That framing also explains why there is a Works With Home Assistant section before the table. The author recommends checking the Home Assistant integrations list and choosing from the list of partner brands when buying new products to avoid unnecessary aggravation, and the closing line of that section is a slogan rather than a fact: support the vendors that support us.

So the list is not only negative. It is paired with a positive recommendation of a certification programme, and the two together make the position coherent: buy from companies that work with the local ecosystem, and if you already own something from a company that does not, this is your reference for what went wrong.

## The table format, and why the entries are one line each

The Hall of Shame table has four columns: Device, Reason it's garbage, Date, and Workarounds if any.

The Date column turns the list into a timeline. The entries run from 2023-11-07 for the Chamberlain MyQ controller through 2024-11-15 for the Ariston Remotethermo, 2025-01-20 for Bambu Labs, 2025-04-12 for the Bovac and Neato vacuum, 2025-08-07 for a batch of Aqara and Amysen entries, 2025-12-04 and 2026-02-23 for later ones. So the list is not a static set of opinions, it is a record that grows as vendors make decisions.

The Workarounds column is the most useful one, and it is where the entries stop being complaints and start being advice.

For the Bambu Labs entry the workaround is simply do not upgrade your printer firmware, which follows from the entry's description of a firmware change that prohibits using unauthorized third-party software to control the printer. For the Bovac and Neato vacuum, which lost its cloud service in 2025, the workaround is a community ESPHome project, called a brainslug, that lets you regain local control. For the MyQ Security 3.0 controller, the workaround is to solder an ESP32 or ESP8266 to the button contacts on a remote, with the note that it is hacky but the vendor cannot disable it.

That last one is the clearest statement of the philosophy behind the list. A vendor can lock out its own app, block third-party integrations, and shut down cloud services, and a physical wire soldered to a button contact is outside their reach.

There is also a Reformed section in the table of contents, alongside Works With Home Assistant and Hall of Shame. The list of criteria, the partner programme and the shame hall are the three headings the document is built from, and Reformed is the fourth, which by its name is the section for products that have changed their behaviour and no longer belong on the list. The truncated excerpt available here does not show its contents.

The maintenance conventions are stated in a comment inside the table. Please keep entries to single lines, it keeps table entries from being corrupted when the table is sorted alphabetically, and new entries should go in alphabetical order to make it easier to find as the list grows. There is a doctoc-generated table of contents at the top with a marker saying not to edit it and to re-run doctoc instead, so the document's structure is maintained by a tool rather than by hand.

## The recurring failure: a cloud service ends

The single most common pattern across the entries is a service shutdown, and the two clearest examples are the garage door and the vacuum.

For the Chamberlain MyQ Smart Garage Door Controller with Security 2.0, the entry records that Chamberlain shut off all access to its APIs, and it links the company's press release and a Verge article about the coverage. The workarounds listed are three separate projects: opengarage.io, whose hardware design, firmware, app code and API documents are all publicly available, ratgdo, which is another solution with open source firmware that works with Security+ 2.0 openers, and the Konnected.io smart garage door opener, which is based on ESPHome and has a Home Assistant integration.

That entry is dated 2023-11-07, and the reason for its existence is structural. A garage door opener is a physical thing on your property that has to keep working. If the only way to command it is a server in a data centre, then the server's existence becomes your property's existence. The three workarounds listed all share the property that they do not depend on Chamberlain's servers.

The later Chamberlain entry, for MyQ Security 3.0, is dated 2025-12-04 and describes Chamberlain doubling down, switching to Bluetooth controllers and excluding pairing to any devices not built by them. So the story is not a one-off; it is a direction of travel, and the second entry exists to record that the first workaround no longer covers the newer hardware.

The Bovac and Neato vacuum entry, dated 2025-04-12, records the announced end of the cloud service in 2025 for all the Neato connected products, and notes that after the shutdown the devices cannot be managed or monitored with the app and have to be started manually. A vacuum cleaner that only runs when a server says so is a vacuum cleaner that stops working.

The Generac Mobilelink entry is the recurring-breakage case rather than a shutdown. It describes a long history of Generac breaking logins, the integration developer fixing it, and Generac breaking it again, dated 2026-02-23 with no workaround listed. The Ariston Remotethermo entry is the same shape, describing continuous disconnection of the API, and adds a detail worth noticing: not only did they break the integration, they log you out of their own app for as long as you are trying to use the integration.

That last detail is the reason Belkin is on the list with the note to not buy. Belkin decided to drop the almost or often mandatory cloud functionality of some devices, the whole Wemo range, effectively rendering them unusable, and the entry says it is no longer possible to pair them with HomeKit or Thread.

## Hardware that fails on its own, and the firmware gate

Not every entry is about a vendor decision. Some are about the device itself, and those entries are the ones that make the list useful to somebody who has not yet bought anything.

The Amysen RGBW bulbs are the first example, and the reason is one of manufacturing variance. These WiFi smart bulbs work with Home Assistant with very little issue, however of the ten or so purchased over about three months, only three remained in service, and all the failed ones developed a driver whine, described as a screech really, within a short handful of months. The entry notes the workarounds column is empty, so there is nothing to do but buy fewer of them and expect some to fail.

The Aqara Water Leak Sensor is the entry with the most detail, and it is a good example of a device that is good and bad at once. The minuses: the manual was unclear on how to put it in pairing mode, you press the water droplet icon firmly until the hidden light flashes three times, and after two months of usage it drops out of the Zigbee mesh a lot even though it is five feet from a Smartthings motion detector that has no problems. The pluses: these ship ready to go, with no messing about with the battery, just press the button and you can add it to your Zigbee mesh. The entry ends with the recommendation not to use it.

The Aqara Vibration Sensor has the same mesh problem from the other direction. Even when it was added successfully, it kept falling off the Zigbee mesh even though other devices within ten feet maintained stable connections, and at least it was cheap.

And then there is the Bambu Labs entry, which is a different category again: a firmware change rather than a shutdown. Bambu announced a firmware change that prohibits using unauthorized third-party software to control the printer, which is why the workaround is to not upgrade the printer firmware. The entry links the company's own blog post about the authorisation control system and a Reddit thread on why the change matters.

That entry is the most interesting of the four because it is about a third-party printer, not a smart home device, and because the failure mode is different in kind. Nothing is broken by a server ending; instead, a device that people had integrated with their own tooling is deliberately made hostile to that tooling, and the loss happens at the moment they update.

## How to use this list, and where it stops helping

There are three practical things to do with a document like this, and one place where it stops being useful.

The first is as a buying guide, which is what the author intends. The Works With Home Assistant section is the positive half of the advice, and the recommendation to check the integrations list and the partner brands list before buying is the concrete form of it. The Hall of Shame is the negative half, and a device on it is a device where you are buying a future support ticket.

The second is as a diagnostic. If something you own stopped working, the list may already contain the answer, and the Date column tells you whether the problem is a recent vendor decision or something you have been living with. The Belkin entry, which says simply do not buy, is also a useful record of why a Wemo device stopped pairing with HomeKit.

The third is as a source of workarounds. The table is only useful as a list if the workarounds column stays populated, and for the MyQ and Neato cases it is: open hardware projects, community ESPHome firmware, and in one case a soldered ESP32. That is the part of the list with lasting value, because a workaround outlives the complaint.

Where it stops helping is hardware you already own and cannot replace, and the Ariston Remotethermo entry is the case in point. The entry records continuous API disconnection and no workaround, and the problem is not that the device is bad but that the integration depends on a server the vendor controls and can break at will. For that situation the list tells you the truth and offers nothing else, and the only durable answer is replacing the device or accepting the dependency.

Two smaller observations. The list is explicitly a work in progress in the sense that the author asks for pull requests and issues adding more, so it is incomplete by design and by nature. And it is opinionated in a way that may not match your priorities: an entry about a driver whine and an entry about a vendor locking out third-party software are both disqualifying, but only one of them is a matter of principle rather than a matter of luck. Nobody is going to list every device that might fail, so the list is a record of the ones that did, which makes it a good guide to the shape of the problem rather than a complete inventory of it.

## Conclusion

Use Internet-of-Trash as a buying guide if you run Home Assistant and are about to spend money on hardware, because its value is the pattern it records, a vendor shutting a service down or adding a firmware authorisation gate and a device that worked for years stopping with nothing in your code changing. Do not treat it as a complete inventory, because it is a list of devices its author and contributors happened to hit, and hardware you already own may not be on it at all. Before you buy, check the Works With Home Assistant integrations list and the partner brands list rather than relying on the negative list alone, and if you do own a listed device, look at the Workarounds column first, since the open hardware projects it points to, opengarage.io, ratgdo, the ESPHome brainslug and the Konnected.io opener, are the durable part and will outlive the complaint.

## FAQ

### What is the Internet-of-Trash list?

It is a list of IoT devices to avoid, aimed at Home Assistant users, with the Hall of Shame table giving each device a reason, the date it was listed and any workarounds. The list is a single README with no code, and the last push to the repository was on 2026-08-08.

### What reasons get a device on the list?

Three: a device had an open API and then switched to requiring you to use their application instead of scraping the API directly, devices which are cloud-only, and anything that requires their application and does not document their API so it can be integrated with other ecosystems.

### Why was Chamberlain's MyQ controller listed?

The Security 2.0 entry records that Chamberlain shut off all access to its APIs, and the Security 3.0 entry from 2025-12-04 records that they switched to Bluetooth controllers and exclude pairing to any devices not built by them. The workarounds listed are opengarage.io, ratgdo and the Konnected.io opener, all of which work without Chamberlain's servers.

### What alternatives are suggested when a vendor breaks a device?

For the Neato and Bovac vacuum, an ESPHome community project called a brainslug that regains local control. For the MyQ Security 3.0, soldering an ESP32 or ESP8266 to the button contacts on a remote. For Bambu printers, not upgrading the firmware, since the change prohibits unauthorized third-party software from controlling the printer.

### How are entries maintained?

The README asks for pull requests or issues adding more, with entries kept to single lines so the table is not corrupted when it is sorted alphabetically, and new entries added in alphabetical order. The table of contents is generated by doctoc and is marked as not to be edited by hand.

## Sources

- [Issues](https://github.com/unixorn/internet-of-trash/issues)
- [README](https://github.com/unixorn/internet-of-trash/blob/main/README.md)
- [unixorn/internet-of-trash on GitHub](https://github.com/unixorn/internet-of-trash)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/unixorn-internet-of-trash