"Without Claude too": CleanMyTV fills the last step for a 50k-star hit, 17 days to 340 stars

CleanYourTV used an AI agent to strip ads from TVs and got 52k stars in three weeks, but you had to be able to use Claude Code first. On the third day of the viral wave, a developer named Vivek built a Mac / Windows version that works with a double-click, no command line needed. The upstream author listed it in the official recommendations, and the public happened to already be used to calling CleanYourTV "CleanMyTV." It doesn't have many stars, but you can explain how each one came in. Below: what it got right, where the ceiling is, and whether this approach still works for you.

"Forget Claude"

In 2026 Nov 5, the YouTube channel Tech It Before You Wreck It posted a video titled "Forget Claude - the EASY way to CLEAN UP a SmartTV": forget Claude, there's an easier way to de-ad a smart TV. The description contained only one download link, pointing to wingedpanther/cleanmytv on GitHub. By 11-06, the video had 43,396 views. The same day, the German magazine Computer Bild ran a tutorial titled "CleanYourTV und CleanMyTV."

By then CleanMyTV had been around for 16 days. On the day the repo was created, Oct 20, the author posted "CleanMyTV – Built from CleanYourTV" on Hacker News, which got 1 points and 1 comments. After that he posted nothing else, and the stars have come from others spreading it.

What it is

A free Mac / Windows desktop app. The first line of the README is blunt: "Clean your Android TV over Wi-Fi… You don't need Python, Java, Android Studio or adb."

Usage is three steps: on the TV, press the version number seven times to open developer options, and put the computer and TV on the same Wi-Fi; open CleanMyTV, which finds the TV on the local network by itself, and tap "Allow" on the TV; then click "Clean & launch." Based on your TV model, it picks the packages to disable from the upstream CleanYourTV list (currently covering 371 models and 100 brands), turns off the manufacturer's ads and preinstalled apps, and installs its bundled ad-free home screen, CleanMyTV Home. Nothing is uninstalled, and one "Restore" button reverts everything.

The second paragraph of the README is the credit: "The per-TV cleanup lists come from @gathint's cleanyourtv. Thank you, Gathin." It doesn't gather data itself. It uses the upstream's data entirely, and only adds the layer that lets ordinary people use it.

Where the stars came from

GitHub daily numbers: 10-24 23 → 10-27 55 → 10-28 108 (single-day +53, the highest of the whole period) → 11-01 198 → 11-06 340, forks 20. It gained 185 in the last 7 days, and recently it's adding 22 to 33 a day, still climbing slightly.

The sources break down into roughly three streams:

1. The upstream officially sends traffic its way. CleanYourTV's README says under "Start here": if you don't have a coding agent, check the community apps. Under the FAQ "Can I use it if I don't know how to code?" it lists 4 community apps, and the cleanyour.tv/app page is the same. CleanMyTV is the only Mac / Windows version on that list. The more the upstream spreads to the public, the larger the share of people who can't use Claude Code, and the bigger the slice it gets: 10-28 the upstream gained 4,222 in a day while it gained 53, about 1.3%; 11-06 the upstream gained 616 in a day while it gained 33, about 5%.
2. The name is a common public mix-up. "CleanMyTV" was already how many people referred to CleanYourTV. Techtippr published a piece saying "CleanMyTV OS is the name people often use for the viral project," a name-riding site wrote "people search for clean my tv," and there's a cleanmytv hashtag page on TikTok. When media cover it, headlines often put the two names side by side, and Computer Bild even cited it alongside another guide site, cleanmytv.com. Whether the author deliberately picked this name can't be seen from outside (inference: most likely yes). He openly wrote "Built from CleanYourTV" in his HN launch post, so he didn't hide it.
3. The YouTube and small-media long tail. Besides the 43k-view video above, a video titled "I used AI to de-BLOAT my smartTV (no PROMPT needed)" also links to its download page. The French site it-connect, Indonesia's omghackers, and a Taiwanese Threads user's post about "no paid AI needed" all mention it. The pitch is the same in every case: no AI, no paying, no coding knowledge needed.

The jump on day 10-28: on the same day, the two phone community apps Tellyscope (+183) and cleanyourtv-app (+231) also hit their highest daily gains of the period, all three rising together. The most likely cause is that the upstream put its community app list in a more prominent place that day (inference; no matching commit was found). On X, nobody discussed it for 30 days, and no sign of paid ads or star farming was seen.

Who uses it

The issues show it: by day 17 the numbering had reached #69. TV models include Xiaomi Mi TV P1, TCL 65C8L, Nokia 8010, Chromecast with Google TV, and plenty of small brands from Russia and Eastern Europe. Several issues are written in Russian ("CleanMyTV won't open on MacBook Air M5"). These people mostly wouldn't install the Android SDK, and the upstream's barrier is what keeps them out.

The author replies fast. One user's issue was titled "Thank you for developing CleanMyTV" and mentioned a small problem; the author replied the same day that the new version had fixed it and added his KIVI TV to the known-models table. Another Chromecast user's screensaver problem took three rounds of back-and-forth before the author pinned down which two packages were being disabled. In the release notes for 0.7.1, three fixes each end with "Thanks nasd223."

He didn't stop at just a cleanup list either: in 17 days he shipped 10 versions, from 0.2 to 0.7.1, adding themes, a wallpaper carousel, weather, now-playing, parental lock, a crackling-fireplace screensaver, drag-an-APK-to-install, and re-closing reopened packages after firmware updates. He also tackled the stumbling blocks every beginner hits, like "macOS blocked local network access" and "VPN is on."

Who it's competing with

Two things stand out. First, the slot "desktop app, Mac and Windows, with a GUI" was the first to fill, and it's the only one on the official list. Second, that slot is small. Among community apps, the phone-based Tellyscope has 1,539 stars, growing 7 in the last 865 days, about 4-5x faster. Most people have an Android phone on hand and don't need to fire up a computer. The group it truly owns is iPhone users who also have a PC at home.

Others are starting to copy it too: one user filed an issue on another launcher project saying its code and settings menu are 1:1 lifted from CleanMyTV, with no credit.

Where the ceiling is

Its stars come from channel placement, not its own demand. The upstream's daily growth has already slowed from 10-22's 7,120 to just over 600. Its share keeps growing, but the total has peaked along with the upstream. It's now gaining about 30 a day. If it follows the upstream down, it could land somewhere between 600 and 900 by year-end (estimate). It's free and open source, so there's no money coming in. For the author, the project bought a spot named by Computer Bild, a Discord community, and a group of users willing to test their machines.

If it were you

Grade C: this slot has already been taken, and the TV wave has passed its peak. What's worth copying is the approach: build the "even non-coders can use it" layer on top of a viral open-source project, then get the upstream to list you as an official recommendation. Timed right, you can reliably pull in a few hundred stars, but rarely more.

Who it suits: developers who move fast, have shipped a desktop or mobile app before, and are willing to answer issues daily. If you want a project with real users that you can put on your resume, not an overnight 10K-star hit.

Who it doesn't suit: anyone wanting to build yet another TV ad-removal tool (the desktop side already has one, and the phone side has two with over 1,000 stars each); anyone hoping for several thousand stars (the biggest community app has only 1,600); anyone unwilling to send PRs to the upstream author and credit them (your traffic almost entirely depends on him).

How to do it:

• Watch the GitHub weekly trending and trendshift: when a project that charts requires installing Python, installing an SDK, or running an AI agent, but the end user is an ordinary consumer, that's a gap. CleanYourTV blew up 10-19, CleanMyTV built a library 10-20; CleanYourTV-Manager, which came out 5 days later, never made the official list.
• Build only the easy-entry layer, use the upstream's data and core logic directly, and state the source and thank the original author in the first paragraph of the README.
• Pick a platform by checking which slot is empty: this time the phone version got 4-5x the stars of the desktop one. Build for whichever device your users already have.
• Take the install experience all the way: how to get past the unsigned-package warning, how to grant permissions, and why a connection fails, each explained in plain language inside the UI.

How to get users: once it's live, open PRs or discussions on the upstream the same day, and apply for the community-apps list, which is the main channel. Lean into the names people actually say and misspell. Then find YouTube creators who make "how to do XX without AI" videos; that angle alone is a video topic. Don't count on HN or X: neither drove real traffic here.

When to get in early: within 3 to 5 days after the upstream goes viral. Once the upstream README's recommendation list fills up, latecomers only get scraps.

What would show this read was wrong: (1) the upstream ships its own official GUI or phone version, which cuts off the community-app traffic; (2) the next mass-market viral project spawns several "no-code versions" that all stall at a few dozen stars, which would mean CleanYourTV's media wave was unusually large and this approach isn't repeatable; (3) conversely, if someone gets several thousand stars by making a no-code version of a viral project, it means this slot is bigger than estimated, so move earlier and choose mobile.