TizenTubeCobalt is a vendored Chromium OS checkout with a streaming app on top
Experience TizenTube on other devices that are not Tizen.
At a glance
- What is it?
- reisxd/TizenTubeCobalt is a BSD licensed Cobalt-based client that removes ads from streaming video and adds SponsorBlock, DeArrow, and playback speed control. The repository is not a small application: the tree carries the Chromium OS source directories, three build systems, five Cobalt container definitions, and three agent directories.
- Who is it for?
- This is a reasonable way to run an ad-free streaming client on a Google TV or other non-Tizen device, and the features it adds are the open ones: ad removal, SponsorBlock segments, DeArrow retitling, and speed control. The ad removal is worth being plain about, since the content is already free to watch and the interruption is the cost being removed, not a paywall being defeated.
- Can I use it commercially?
- Yes. BSD-3-Clause is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 30 days ago.
- What is it written in?
- GitHub does not report a main language for this repository.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- reisxd/TizenTubeCobalt3,574 ★ · +442
| Repository | 2026-09-24 | 2026-11-15 |
|---|---|---|
| reisxd/TizenTubeCobalt | 3.1k | 3.6k |
442 Stars in the last 52 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
The tree is a Chromium OS checkout with an app layered on top
The top-level directory listing is the first fact worth reading, and it is not the listing of a streaming client.
It contains `chrome/`, `chromeos/`, `chromecast/`, `cc/`, `content/`, `components/`, `crypto/`, `dbus/`, `device/`, `net`-adjacent build infrastructure, `ash/`, `base/`, `apps/`, `build/`, `buildtools/`, `build_overrides/`, and `clank`. Alongside them sit `DEPS`, `BUILD.gn`, `.gn`, `DIR_METADATA`, `OWNERS`, `WATCHLISTS`, `codereview.settings`, and three presubmit scripts named `PRESUBMIT.py`, `PRESUBMIT_test.py`, and `PRESUBMIT_test_mocks.py`.
That is the Cobalt runtime's own source tree, checked in. The README says so directly: Cobalt is a lightweight, cross-platform application container and runtime for HTML5-based apps, originally developed by Google for embedded and resource-constrained devices such as smart TVs, set-top boxes, and game consoles.
So this repository is a vendored copy of an entire Chromium-derived operating system, with a single application added to it and a container definition to build it. The application is `cobalt/`, and the container recipes are under `cobalt/docker/`.
The consequence for anyone evaluating this is scale. There is no small build here. Getting a binary out of it means building a Cobalt image, not running a script.
Five style and lint configurations for three languages sit in one root
The root carries a formatting or linting configuration for each language the tree contains, and there are five of them.
For C++ there are three: `.clang-format`, `.clang-tidy`, and `CPPLINT.cfg`. For Rust there is `.rustfmt.toml`. For Python there are two, `.pylintrc` and `.yapfignore`, which is a yapf ignore list rather than a full configuration.
Two of the entries are hints about how this is actually worked on. `.clangd` is a language-server configuration for clangd, which means the C++ side is meant to be edited in an editor with completion rather than in a bare shell. `.pre-commit-config.yaml` is the only file that ties the formatters together, and it is the one file in the list that would actually run.
There is also a `.git-blame-ignore-revs` file, which lists commits whose changes are formatting-only so that `git blame` attributes a line to the last substantive change rather than to a whitespace pass. Its presence in a vendored tree is the clearest signal that the formatting tools above have been run across the checkout at least once.
Six more root files describe ownership rather than code: `ATL_OWNERS`, `CRYPTO_OWNERS`, `SECURITY_OWNERS`, `AUTHORS`, `.mailmap`, and `CODE_OF_CONDUCT.md`, all carried over from the upstream tree.
Three agent directories sit in a Chromium tree, and the project records no primary language
Three directories at the root are named for tooling that automates work rather than for the application itself.
There is `.gemini/`, `.agent/`, and a top-level `agents/` directory. So three different conventions for agent instruction files exist in one repository, in a tree that is otherwise inherited from Chromium.
That is a small signal but a real one about the fork. When a project is a vendored upstream checkout, the question is always where the line between upstream and local is. Three agent directories with three different names do not make that line obvious.
The language field for the repository records nothing at all, which follows from the same cause. With the C++ of Chromium dominating the tree, the tooling that classifies the project cannot attribute a primary language to the fork, so the number is simply absent.
The same pattern shows up in the documentation set. There is a `codelabs/` directory, a `docs/` directory, and `CODE_OF_CONDUCT.md` plus `CONTRIBUTING.md` at the root. The README itself is the only document in the repository that describes the application, and it is under two hundred words long.
Five container recipes, three platform pins, and one variable that retags all five
The compose file builds five images, one per Cobalt configuration:
linux:
build:
context: cobalt/docker/linux
image: "${DOCKER_COMPOSE_IMAGE:-linux}"
platform: linux/amd64The five are `linux`, `raspi`, `rdk`, `unittest`, and `webtest`, each with its own build context under `cobalt/docker/`.
Three of them declare `platform: linux/amd64`. Two do not:
unittest:
build:
context: cobalt/docker/unittest
image: "${DOCKER_COMPOSE_IMAGE:-unittest}"So the test images build for whatever architecture the host is, while the three target images are pinned. On an arm64 machine that produces test binaries and target binaries for different processors, which is the kind of mismatch that produces a confusing failure rather than a clear one.
The second detail is the image name. All five services read the same environment variable with a different default, so a build with `DOCKER_COMPOSE_IMAGE` set produces five images carrying the identical tag and the last one written wins in the local image store. The variable looks like a versioning mechanism and behaves like a collision.
There is nothing else in the file. No ports, no volumes, no networks.
One of the five feature bullets is a request for more features
The features section has five entries, and the fifth is not a feature.
Four do something. The ad blocker removes ads from streaming sites. SponsorBlock support skips sponsored segments automatically. Video speed control adjusts playback speed. And DeArrow support removes clickbait and misleading video titles, which is a separate public project that rewrites titles rather than a local heuristic.
The fifth asks for requests through the issue tracker.
So a headline count of five is really four capabilities plus a placeholder, and the four are all open-source services or client-side behaviour rather than proprietary advantage. Two of them, SponsorBlock and DeArrow, are the two public projects you can also point any other client at.
That is a reasonable thing for this project to be. It is worth being clear that the delta against another client is the runtime underneath it, not the feature list.
Distribution is a store channel code, with sideloading named as the alternative
The download section names two routes.
The first is a store channel: an AFTVNews code, `6366500`, which is the numeric identifier a Tizen television app store uses to resolve a channel's application list. The second is the release page, linked as the latest release.
The install section then describes three ways to get it onto a device: sideload or install using a file manager, ADB, or a platform-specific method. And it recommends, for a better experience, using the app on a Google TV certified device, with a link to a third-party Android TV guide.
So the intended path is a television that can install from a store channel, with manual installation as the fallback for a device that cannot.
That recommendation is worth taking seriously for a specific reason. This client is built on a Chromium OS runtime rather than on the platform's own web stack, so it depends on a browser engine being present on the device. A Google TV certified device is one where that engine exists in a known state. A device where it is absent or heavily modified is where the black screen or the crash would come from.
The site the repository points at as its homepage does not appear anywhere in the README, which links only the release page, the issue tracker, and three community channels on Discord, Telegram, and Matrix.
The newest release tag is v2.0.2 from May 2026, eleven days before the last push
Three releases are visible, all from May 2026. v2.0.2 was published on 2026-05-17, v2.0.1 on 2026-05-10, and v2.0.0 on 2026-05-09.
The first two are one week and one day apart, and the third is one day before the second. That is the shape of a project that shipped a 2.0 line in three quick steps rather than one planned release.
The last push to the main branch was on 2026-05-28, eleven days after the newest tag. So there is work on the branch that no release covers, and the newest tag is not the newest code.
The tags themselves are worth noting for anyone writing automation. All three carry a `v` prefix and are plain two-component versions, and the release names duplicate the tag with the project name in front, so a name reads TizenTube Cobalt v2.0.2.
The homepage is a separate domain from the repository, and the description says the project lets you experience TizenTube on devices that are not Tizen. So the lineage is a Tizen television client ported to a different runtime, which is what the Cobalt description in the README points at.
Editorial conclusion
This is a reasonable way to run an ad-free streaming client on a Google TV or other non-Tizen device, and the features it adds are the open ones: ad removal, SponsorBlock segments, DeArrow retitling, and speed control. The ad removal is worth being plain about, since the content is already free to watch and the interruption is the cost being removed, not a paywall being defeated. Two things to weigh before you clone it. The repository is a Chromium OS checkout, so the build is a Chromium build and nothing about it is small. And the release tags stopped in May 2026, with the branch moving a few days after the newest one, so pin a commit rather than assuming the newest tag is the newest code.
Frequently asked questions
What is TizenTubeCobalt?
It is a BSD licensed streaming client based on Cobalt, the Chromium-derived application container and runtime originally developed by Google for embedded and resource-constrained devices such as smart TVs and set-top boxes. Its features are ad blocking, SponsorBlock segment skipping, DeArrow retitling, and video speed control.
How do I install TizenTubeCobalt?
The README gives three steps: download the latest release, sideload or install the app on the device using a file manager, ADB, or a platform-specific method, then open the app. It names an AFTVNews code of 6366500 for store distribution and recommends a Google TV certified device for a better experience.
What is in the TizenTubeCobalt repository?
A full Chromium OS and Cobalt checkout rather than a small application. The top level carries chrome/, chromeos/, chromecast/, cc/, content/, components/, crypto/, ash/, base/, build/, buildtools/, clank, and cobalt/ directories, along with DEPS, BUILD.gn, OWNERS, DIR_METADATA, and three presubmit scripts.
What containers does the TizenTubeCobalt compose file build?
Five, from contexts under cobalt/docker/: linux, raspi, rdk, unittest, and webtest. The linux, raspi, and rdk services set platform: linux/amd64, while unittest and webtest do not. Each service names its image with the same DOCKER_COMPOSE_IMAGE variable and a different default value.
What are the latest TizenTubeCobalt releases?
v2.0.2 published on 2026-05-17, v2.0.1 on 2026-05-10, and v2.0.0 on 2026-05-09. The main branch was last pushed on 2026-05-28, so the branch has moved past the newest tag.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/reisxd-tizentubecobalt)Star history chart for your README
[](https://hysenlabs.com/projects/reisxd-tizentubecobalt)Was this useful?