About ClipVault

Why it exists

Saving a video usually means walking through a site that is 60% advertising, three fake download buttons, one installer you did not ask for, and finally a watermark across the top of the result. Or it means installing a desktop app, keeping it updated, and hoping the platform upstream did not break it that week.

This is the version of that with nothing extra in the way: one box for the link, a list of the formats the source actually offers, and the file. No watermark, no redirect chain, no account, no installer.

What it supports

Nine platforms: YouTube, TikTok, Instagram, X (Twitter), Facebook, Niconico, SoundCloud, Dailymotion and Streamable. For video that means every MP4 variant the platform exposes, up to 4K where it exists; for audio it means MP3 or M4A, which is also the quickest route if the sound is all you need.

Each of those platforms gets its own page, explaining what that one site exposes and what the download actually returns — the unsupported cases included.

How it works under the hood

The browser does almost nothing. When you paste a link, the server asks the platform what formats are available, then hands you that list. When you pick one, the server downloads the stream with yt-dlp; if the platform stores video and audio as two separate streams — YouTube at high resolutions usually does — they are merged with ffmpeg before anything is sent to you. The finished file is streamed back to your browser and then deleted from the server, in that order, in the same request.

Nothing is stored. There is no library of past downloads, no copy kept for the next person, no database entry saying what you fetched. The temporary directory for a job exists for the length of that download and is then removed.

Because all the work happens on one machine, there are only two limits, and they have nothing to do with how many files you download: two at a time across the whole service, one at a time per visitor. Download as many as you like — just not all at once.

Who it is for

Mostly people saving their own work, or something they have permission to keep: someone who uploaded a clip and lost the original, a teacher archiving a lecture before it goes private, a musician grabbing their own track for a demo reel, a journalist keeping a copy of a post that may be deleted. Also people on a slow or capped connection who want the file rather than a stream, and anyone who just wants to watch something on a plane.

What it is not for: redistributing other people's work, pulling down a paid course, or bulk-harvesting a catalogue. That is written into the terms, and it is the honest limit of what a tool like this should be used for.

Who runs it

One person, one server, no company behind it and no investors to report to. The site is built with Next.js, yt-dlp and ffmpeg, and the hosting is paid for by the advertising on the page — which is the entire reason those ads are there. If a platform changes its internals and a download stops working, it gets fixed when it gets noticed.