Home Automation

Home Assistant92.0k★

Home Assistant runs your smart home on your own hardware instead of somebody else's cloud. Lights, locks, cameras and sensors land in one app and one automation engine, across more than 1,400 integrations. Apache 2.0, free, no feature tiers. Everything runs locally by default, which is the whole point. Home Assistant OS on a dedicated box handles its own updates and apps (the new name for add-ons). The Docker container leaves updates to you and can't run apps, so no out-of-the-box Thread or Z-Wave. Budget a weekend. Hardware is the real bill: Green is $199 preinstalled, plus $49 for a ZBT-2 (Zigbee or Thread) or $69 for a ZWA-2 (Z-Wave). Home Assistant Cloud, $6.50/month or $65/year, adds remote access, offsite backups, Alexa and Google Home; it becomes Home Assistant Link in December. Solo and households: free plus a Green. Small offices: free. Larger teams: wrong tool. The catch: integrations break when vendors move. 2026.10 removed SolarEdge Local and Melnor RainCloud after their makers killed the APIs. A critical March 2026 advisory also caught some apps exposing unauthenticated endpoints, even a full shell, to the local network until a Supervisor patch. Keep it updated.

CleanYourTV52.9k★

CleanYourTV strips the ads and manufacturer junk off an Android TV or Google TV and swaps the home screen for a clean, Apple TV style launcher. It's free and MIT-licensed. The installer is mostly a prompt: paste PROMPT.md into Claude Code or another coding assistant with shell access, name your TV model, and it finds the TV on Wi-Fi, backs up, debloats and builds the launcher. Setup is turning on USB debugging and one "allow this computer" tap. Nothing is rooted or uninstalled: packages get switched off with pm disable-user, and a restore script turns them back on. You get a small Python ADB client, per-model debloat lists from other owners, and the launcher as Java source you build yourself, with a JDK and the Android SDK or inside Docker. No prebuilt APKs, good call. Fire TV gets the debloat half only; Samsung and LG don't run Android. Solo, household or small office: free. The catch: you're giving an AI agent a shell on a box in your living room, through a debug port that stays open until you close it. Read the list before it runs, keep the restore script, then switch debugging off and confirm with the bundled port check. A few shared lists also disable the TV's own updater, so look for that line in yours.

aerial-tv10.2k★

Aerial is a tvOS-style home screen for the Xiaomi, TCL and other TVs sold in China, whose stock launchers are wall-to-wall show recommendations and membership badges. You get a clock, your apps in a grid, Bing's photo of the day or your own wallpaper, and Apple TV's aerial films as the screensaver. No ads, no feeds. It's MIT licensed and free. Install takes ADB and a few minutes: flip on debugging with the brand-specific button combo from the README, then paste a prompt into Claude Code or Codex, run the one-line curl | bash installer, or adb install the APK from GitHub releases. Solo or a household, it's free per TV. Google TV owners have Free My TV for the same job. LG, Samsung and Huawei sets aren't supported. The catch: it stays in charge by fighting the firmware. It holds secure-settings access, an accessibility service, a notification listener used only to restart itself, and an optional switch that turns ADB back on whenever Xiaomi shuts it off. The APKs are hand-built, five releases landed in a single day, and a firmware update can undo the takeover.

integration7.9k★

HACS is the app store Home Assistant never shipped with. Home Assistant is the open source brain a lot of people run their smart homes on, and the official integrations only cover so much. HACS gives you a UI to find, install, and update the thousands of community-built add-ons that fill the gaps. Free, MIT licensed, community run. Without it, installing a custom component means copying files into the right folder by hand and remembering to do it again every time there's an update. HACS turns that into a few clicks: browse, install, update, remove, with links straight to each project's repo and issue tracker. It's the difference between maintaining a smart home and fighting one. Run Home Assistant and ever wanted a device or dashboard it didn't support out of the box? This is the first thing to install after Home Assistant itself. Solo tinkerers and whole households alike, it's free for everyone and no version costs money. The catch is the same as any unofficial store: you're installing code from strangers. HACS doesn't vet what's in those repos, it just makes them easy to get. Most community components are fine, but you own the trust decision, and an abandoned add-on can break after a Home Assistant core update with nobody on the hook to fix it.

free-my-tv5.9k★

Free My TV swaps the ad-stuffed Google TV home screen for a clean one: your streaming apps on a shelf, hourly trending picks from apps you already have, weather, and a look that darkens at sunset. No sponsored rows, no autoplay promos. It's MIT licensed and free. LG and Samsung TVs get the same design as an app beside the stock home screen. Setup is a real project. You turn on developer mode and USB debugging, then build the APK yourself with a JDK and the Android SDK (there's no prebuilt Google TV download), or let Claude Code do the building and installing for you. Trending picks need a free TMDB key. Nothing is uninstalled: the stock launcher and promo apps are switched off, and one button brings them back. Solo or a household, it's free per TV. For China-market sets that ship without Google TV, look at Aerial. The catch: the switch runs through the TV's own debugging port, so debugging stays on and the app holds shell-level control of your TV. The project is days old, run by one maintainer, and the home screen shows Binance crypto prices unless you turn them off.

skylight3.4k★

Skylight projects the aircraft flying over your house onto your ceiling in real time, with the actual sky behind them: sun, moon, bright stars, and the ISS. It reads ADS-B signals from a cheap software-defined radio, enriches each plane with its airline and route, and renders the whole thing as a live ceiling display. MIT licensed and free. This is a hardware project, so the cost is parts and patience, not software. You need an RTL-SDR radio with an antenna, a Raspberry Pi 5, and a projector pointed up. You can try it with no radio at all using a free public flight API, which is the smart way to see if you like it before buying anything. The docs and deploy scripts are solid, but configuring the Pi and aligning the projector is real setup work. There is no team-size calculus here. This is a personal project, an art installation, and a genuinely fun way to learn how ADS-B and software-defined radio actually work. If you have ever wondered what is flying overhead, build it. If you want a practical tool for shipping software, this is not one, and that is exactly the point. The catch is the hardware tax and the radio reality. Local ADS-B reception depends on your antenna and line of sight, so a bad spot means spotty planes. A commercial ready-made kit is in the works, which is fine, but the free path is the source code and your own parts bin.

cafe-hass1.7k★

cafe-hass is a drag-and-drop editor for Home Assistant automations. Instead of hand-writing YAML, you build your logic as a flowchart and it transpiles down to native Home Assistant automations. If you have ever stared at indentation errors trying to get a motion sensor to do the right thing, this is the visual layer that fixes that. MIT, free, installs through HACS. Because it outputs standard Home Assistant YAML, there is no separate engine running, the automations are native once generated, which means you are not adding another always-on service to babysit the way you would with Node-RED. It lives entirely inside your self-hosted Home Assistant. Setup is trivial if you already run HACS. This is for Home Assistant users who want visual automation building without standing up Node-RED alongside it. Solo home setups, which is basically everyone here, it is free, full stop. The flip side: it only matters if you run Home Assistant, and because it compiles to YAML, deeply custom logic can still bump into Home Assistant's native automation limits. The catch is the niche. cafe-hass solves a real annoyance, and it solves it for exactly one audience: people already running Home Assistant. Outside that world it has no reason to exist. Inside it, it is one of the nicer ways to stop fighting YAML.

esp32cam-rtsp1.2k★

esp32cam-rtsp turns a cheap ESP32-CAM board into a network camera. Flash this firmware onto the chip and it serves a live video stream over RTSP (the protocol VLC and most CCTV recorders speak), plus a browser-viewable Motion JPEG feed and still-image snapshots, all configured through a built-in web page. It supports a wide range of ESP32-CAM variants, from the common AI Thinker board to M5Stack and Seeed XIAO modules. Running it means real hardware work. You compile and upload with PlatformIO, put the board into download mode to flash it, then connect to its setup access point to enter WiFi and camera settings. From there it exposes brightness, contrast, exposure, and gain controls. Once flashed it just runs, but the initial bring-up assumes you're comfortable with embedded toolchains and finicky USB flashing. Solo tinkerers and small teams building a DIY camera or a vision project: this is a clean, well-supported way to do it for the price of a board. Large teams: fine for prototypes, not for anything needing fleet management or security guarantees. The catch: the repo declares no license, so "open source" is doing a lot of work here. The code is public and free to use, but without a stated license your legal right to redistribute or build a product on it is unclear. For a weekend project that's irrelevant. For anything commercial, ask the maintainer before you ship.

gpt-home652★

Turns a Raspberry Pi into a voice assistant powered by ChatGPT. Plug in a microphone and speaker, run the install script, and you've got a smart speaker that uses OpenAI's API instead of Alexa or Google Home. Say something, it sends it to GPT, speaks the response back. The setup is a Docker container on a Raspberry Pi (3B+ or 4 recommended). It handles wake word detection, speech-to-text, GPT completion, and text-to-speech in a pipeline. Supports Home Assistant integration so you can control smart home devices by voice. Completely free software. Your cost is hardware (a Pi you probably already have) and OpenAI API usage. At casual home use (maybe 20-30 queries a day), expect $1-3/mo in API costs. The catch: this is a hobby project. It works, but don't expect Alexa-level reliability or polish. Wake word detection isn't perfect. Latency is noticeable; you're making a round trip to OpenAI's API for every query. There's no offline fallback. And if OpenAI changes their API or pricing, you're dependent on the maintainer updating. For a more robust self-hosted voice assistant, Home Assistant's own voice pipeline is more mature.

Anima637★

Anima is a local brain for a Xiaomi smart home. Instead of writing rigid "if motion, then light" rules, you run an LLM agent that figures out what your devices are and decides what to do based on context and learned habits. Apache-2.0 licensed and self-hosted, you bring your own LLM API key. Setup is real work. You need the Python and Node stack running, an MQTT broker, an LLM key, and physical Xiaomi or MIoT hardware with the device tokens and local IPs sorted out. Once running, it discovers devices, plans actions through the LLM, and keeps an evidence-based memory of your preferences over time. There's a React dashboard and a full REST API. The obvious comparison is Home Assistant, which is more mature, hardware-agnostic, and has years of community behind it. Anima's bet is that an LLM agent beats hand-written automations. If you're deep in Xiaomi gear and like living on the edge, try it. Everyone else: Home Assistant, possibly with an LLM layer bolted on. The catch: it's days old, Xiaomi-only, and self-described as early-stage. The cloud login handles discovery, but actual control still needs each device reachable by local IP and token. Promising direction, very green.