How Trove works

One box, every platform — how a query travels from your browser to your server and back as a page of results you can open with a single click.

The journey of a single query

Trove is a search front end, not a video host. Nothing you find is stored by the platforms on your behalf; the software simply asks eight video sites the same question at once and shows you what comes back. Here is the full path, step by step.

Search or paste a link Type a query into the search box and press enter. Or paste a direct video URL instead. The provider chips below the box let you decide which platforms take part in the search.
Your server fans out to the platforms Trove forwards the query to every selected platform in parallel, collects the responses, merges them into a single grid and removes duplicates, so the same video does not appear twice because it was indexed by two sources.
Click a result Clicking a result starts the video on the original platform in a new tab, which is always the friendliest option for the creator. When downloads are enabled you get an explicit choice first: Open launches the video, and Download asks which format you want.
Your server resolves the available formats The backend runs yt-dlp against the video and asks the platform what it actually offers: video, audio or image, each with quality and size labels so you can choose between, say, a small 360p file and a large 1080p one.
The file streams through your server The download runs server-side and streams to your browser with a live progress bar and a cancel button, so a long file never blocks the rest of the site. The output is normalised so it plays in QuickTime and VLC without you hunting for extra codecs.

Search features

Provider chips. Each chip is a toggle. Leave them all on to cast the widest net, or switch most of them off when you only care about one or two sources — a handy way to keep a search fast and the results focused.

Trending. The home page shows what is popular right now with period tabs for Today, This Week and This Month, so you can see what people are watching over a window that suits you rather than a single frozen snapshot.

Popular Searches. A public section built from aggregated, lowercased query terms that other people have run on this server. It is a discovery aid: you get ideas for what to search next, and no individual query is shown next to a name or profile, because there are none.

Direct-link mode

You do not have to search at all. Paste the URL of any supported video straight into the box and Trove resolves it, shows you the formats that are available, and hands you the same Open or Download choice. Because downloads are handled by yt-dlp, this works for a great many sites well beyond the eight that the search covers. See the supported platforms page for the distinction between search and download coverage.

The privacy model

When you search from your own instance, the query leaves your server — not your personal browser. The video platforms therefore see a request from your machine rather than a request from you with your cookies, your history and your recommendations attached. There is no account, no sign-up and no profile, so searches are not tied to an identity. Our Privacy Policy spells out exactly what is stored where, in plain language.

What self-hosted means here

Trove runs on hardware you control: a FastAPI backend, a static web interface and companion iOS and Android apps built with Kotlin Multiplatform, all pointed at your own address. You decide what data exists, how long it stays and who can reach the server. If you want to wipe the stats files and start fresh, you can — they are ordinary files on your disk. The trade-off is honest: you are responsible for keeping the server and its yt-dlp installation up to date.

Still have questions? The FAQ covers legality, mobile apps and troubleshooting, and the contact page reaches the operator directly.

Last updated: October 2026