# ViewTube self-hosting and YouTube data flow > ViewTube is a self-hosted YouTube frontend built with Nuxt and Nest, using scraped website data rather than the official YouTube API. Its main attraction is a separate account and subscription experience, but operators accept the fragility and hosting responsibilities that come with web scraping. **ViewTube/viewtube** — ▶️ ViewTube: The open source, privacy-conscious way to enjoy your favorite YouTube content. Docs: https://viewtube.wiki - Repository: https://github.com/ViewTube/viewtube - Website: https://viewtube.io - Stars: 1,475 · Forks: 91 - Language: TypeScript - License: AGPL-3.0 - Published: 2026-10-16 · Updated: 2026-10-16 · Language: en - Canonical page: https://hysenlabs.com/projects/viewtube-viewtube ## Why ViewTube replaces the standard YouTube interface ViewTube targets people who regularly use YouTube but want a frontend that they can host themselves and manage separately from their YouTube identity. According to the README, it provides search, video playback, comments, playlists, channel subscriptions, and push notifications for new uploads without presenting the standard YouTube frontend. The goal is not a downloadable video archive. It is an alternative browsing and subscription interface for content still obtained from YouTube. That distinction matters for adoption. ViewTube does not remove the need to access YouTube content or guarantee uninterrupted availability of it. It changes the application layer placed in front of that content. The project is most relevant to a self-hosting user who wants control over the web interface, a local account system separate from YouTube, dark and light themes, and a touch-oriented player. The README also lists playback controls such as loop, speed, autoplay, volume, and gestures, plus integrated SponsorBlock support. These features make it more than a bare search page. Still, the repository description and README present no evidence of offline downloads, background audio, or a fully independent content catalog. ViewTube should therefore be evaluated as a hosted frontend, not as a replacement media repository. ## How ViewTube retrieves YouTube data without the official API The central architectural decision is documented directly in the README: ViewTube does not use the official YouTube API. Instead, it retrieves data by scraping the YouTube website through custom tooling and two named libraries, `YouTube.js` and `node-ytpl`. Search results, channel information, playlists, comments, subscriptions, and playback data therefore depend on the behavior of the public website and the extraction libraries rather than on a stable API contract published for application developers. The repository layout shows a `pnpm` workspace divided into `client`, `server`, and `shared`. The root `package.json` identifies Nuxt as the client-side framework context through its scripts, while the README identifies both Nuxt and Nest as the main technologies. This separation gives the application distinct client and server responsibilities, with shared code available across workspace packages. The exact internal request routing and scraper control flow are not documented in the README, so a more detailed call sequence cannot be confirmed. This design explains both ViewTube's value and its main operational risk. Avoiding the official API can remove API quota and approval concerns, but it makes the project sensitive to website changes. The README does not document a compatibility fallback, retry policy, or behavior for a failed scrape. An operator should not assume that every upstream change will be handled transparently. ## MongoDB, Redis, and the ViewTube service topology The repository's `docker-compose.yml` defines one application service and two backing services. `viewtube` is built from the repository's `Dockerfile`, exposes port 8066, and mounts `./data` at `/data`. It depends on both `viewtube-db` and `viewtube-redis`. The database service uses the `mongo` image and maps host port 27018 to MongoDB's port 27017. The Redis service uses the `redis` image but publishes no host port. All three services share the `viewtube-local` network. ViewTube receives the internal service names through `VIEWTUBE_DATABASE_HOST`, `VIEWTUBE_REDIS_HOST`, and `VIEWTUBE_ADMIN_USER`. The compose file does not show passwords, a configured host URL, a reverse proxy, or a TLS terminator. That makes it useful as repository-level service topology, but it does not by itself establish a production-ready public deployment. The documentation site is the proper place to check the complete setup because the README sends installation questions there rather than reproducing them. The `Dockerfile` builds with `node:26-bookworm`, installs production dependencies after compilation, and starts the compiled Nest server. It also copies the compiled client output and shared output into the runtime image. Its health check requests the application root and permits a 60-second startup period. This is a meaningful deployment detail: the container is designed around the server process while shipping the client and shared packages as built artifacts. ## Installing ViewTube and reaching the first usable screen The README does not provide a self-contained installation procedure. It directs administrators to `https://viewtube.wiki/installation` and developers to `https://viewtube.wiki/development`. Because the README does not document the prerequisite setup, administrator credentials, first-login flow, or first search, a complete first-run sequence cannot be confirmed. Inventing those steps would be less useful than identifying the boundary. For contributors, the root `package.json` declares Node.js `>=26.7`, pnpm `>=11.23`, and `packageManager` `pnpm@11.24.0`. It defines a development service script that starts the compose file: ```bash pnpm run serve:env ``` This command is documented by the repository and invokes `docker compose -f docker-compose.dev.yml up -d`. It starts the development environment in the background, but the excerpt does not explain what follows or where the developer UI should appear. The separate `start`, `serve:server`, `serve:client`, and `build` scripts indicate distinct build and runtime paths, yet their correct order is not documented here. For a first real user workflow after installation, the README supports a clear use case: open the hosted interface, search for a video, play it, subscribe to its channel, and then inspect that channel's latest uploads. It also documents comments, playlists, push notifications, and SponsorBlock. The README does not state whether subscriptions synchronize with YouTube, so ViewTube accounts and channels should be treated as separate state within this application. ## Scraping compatibility is ViewTube's principal failure boundary ViewTube is the wrong tool when YouTube website compatibility is a non-negotiable requirement. The README states that data comes from scraping and identifies libraries used for that work, but it provides no uptime guarantee for extraction, no formal YouTube compatibility policy, and no fallback when selectors or response formats change. A working playback page today therefore does not prove that the same breadth of search, comments, playlists, and subscriptions will remain available after an upstream change. This limitation is more consequential than the UI feature list suggests. Ads or tracking may be absent from the ViewTube frontend, but content retrieval still depends on YouTube. If a video cannot be resolved, a channel cannot be scraped, or the upstream response format changes, users may encounter an empty result or failed page. The documentation excerpt does not define the visible error state, retry limits, or fallback player behavior. Self-hosting introduces a second boundary. The operator is responsible for the database, Redis, application container, stored data, updates, and public exposure. The compose file publishes MongoDB on host port 27018, which deserves explicit review before deployment on an untrusted network. The README does not document authentication configuration for that port or production hardening beyond the service topology, so copying the compose file onto a public host without checking those details would be an assumption rather than a documented deployment practice. ## Official YouTube API frontends versus ViewTube scraping The most direct alternative is a frontend built against the official YouTube API. That approach changes the dependency rather than removing it. An API-based client relies on documented endpoints, quotas, access rules, and the terms associated with YouTube's developer platform. ViewTube instead extracts data from the website through `YouTube.js`, `node-ytpl`, and custom tools. The official API path can be more suitable when an organization requires a supported integration contract or already has approved API access. ViewTube can be more relevant when the goal is self-hosting the frontend without using that API. The README does not provide a feature-by-feature comparison, supported API plans, or quota information, so neither approach can be claimed to have broader content coverage or greater reliability without further evidence. For evaluation purposes, the choice is concrete. Use ViewTube when scraping-based operation is acceptable. Evaluate an official API frontend when API access and its operational constraints are acceptable. ViewTube's separate account and subscription features do not erase that architectural distinction. ## ViewTube upgrades, AGPL-3.0, and repository activity The repository was last pushed on 2026-10-15 and is not archived. That supports the limited conclusion that repository activity is current, although the README's listed releases stop at `v0.17.0` from 2024-08-27. The root `package.json` reports version `0.17.1`, so the package version and the release list are not aligned in the available repository data. An adopter should inspect `CHANGELOG.md` and release artifacts rather than assume which version is intended for deployment. Upgrade work also depends on the self-hosted services. Application changes may need matching MongoDB or Redis handling, and the compose file uses unpinned tags, `mongo` and `redis`. Those tags do not identify a reproducible dependency version. The project provides a Renovate configuration and tests, but the README does not document a supported upgrade procedure or database migration policy. The project is licensed under `AGPL-3.0`, while `package.json` shortens the identifier to `AGPLv3`. Anyone modifying or distributing ViewTube should review the repository's `LICENSE` and assess how the AGPL terms apply to their deployment and distribution model. This is a compliance review point, not a conclusion that every deployment has the same obligation. ## Conclusion ViewTube fits a technically capable operator who wants to self-host a separate YouTube frontend and accepts that the project depends on scraping rather than an official API. It is a poor fit for users who need a stable commercial service, and operators should first verify the current installation guide at `viewtube.wiki/installation`, the behavior of a fresh deployment on port 8066, and whether the scraping-dependent user flows work in their own environment before importing real data or adding public accounts. ## FAQ ### Does ViewTube use the official YouTube API? No. The README states that ViewTube scrapes the YouTube website using custom tools, `YouTube.js`, and `node-ytpl`. ### How do I install and self-host ViewTube? The README points administrators to `https://viewtube.wiki/installation`. The repository also contains a Dockerfile and a compose topology that exposes ViewTube on port 8066. ### Can ViewTube manage subscriptions separately from YouTube? The README says users can create an account separately from YouTube, subscribe to channels, view latest uploads, and receive push notifications. It does not document synchronization between ViewTube and YouTube subscriptions. ## Sources - [License: AGPL-3.0](https://github.com/ViewTube/viewtube/blob/development/LICENSE) - [Project website](https://viewtube.io) - [README](https://github.com/ViewTube/viewtube/blob/development/README.md) - [Releases](https://github.com/ViewTube/viewtube/releases) - [ViewTube/viewtube on GitHub](https://github.com/ViewTube/viewtube) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/viewtube-viewtube