安装 切换下拉菜单

• 安装压缩版
• 用户脚本新手指南

max / Open Media Grabber

Download video, audio and images from social platforms. Everything happens in your browser — there is no server involved, nothing is tracked, and no link is ever rewritten with an affiliate code.

Source code, issues and full documentation: https://github.com/EfM8caQE9F1v4pFt5x7WrjDI1N2kFqYdydVj/open-media-grabber

Why another downloader

Most all-in-one downloader userscripts work the same way: they send the page URL to a service the author runs, and that service does the extraction. That is where the "human verification" waits, the affiliate links and the extra extension prompts come from — the backend has to be paid for somehow.

None of it is necessary. The platform has already sent the media to your browser; a downloader only has to notice. Keeping the extraction on your machine removes the reason for the ads to exist.

What that means concretely: this script contacts no host except the site you are on and the CDN that site is already streaming from. There is no server behind it. There is nothing to monetise, and no antifeature to declare.

The code here is unminified on purpose. A script that asks for GM_xmlhttpRequest should be one you can read before you trust it. Every external URL in the file belongs to a platform CDN — you can check that yourself with a text search.

What it does

• Reddit — merges the separate video and audio files into one MP4. This is the track most other downloaders silently drop, which is why Reddit videos so often arrive without sound. Galleries are supported.
• X (Twitter) — every MP4 rendition plus the HLS stream; photos at original resolution.
• TikTok — the download copy and the playback copy, both labelled so you can choose; photo-mode posts and the audio track too.
• Instagram and Threads — reels, posts and carousels.
• Facebook — SD and HD renditions.
• YouTube — subtitles and thumbnails reliably; video only when YouTube exposes a plain URL. See the limitation below.
• Any other site — whatever the page fetches or has loaded in a player.

Separate video and audio tracks are combined without re-encoding. The merge rewrites the container header only, so it runs at copy speed, uses a few hundred kilobytes of memory regardless of file size, and loses no quality. There is no bundled ffmpeg.

On Chromium the download streams straight to disk, so a multi-gigabyte file does not have to fit in memory.

Honest limitations

• YouTube. YouTube has moved its web player to SABR, where the page no longer receives a fetchable media URL. When that applies, the panel says so and points you at yt-dlp instead of pretending. Subtitles and thumbnails still work.
• DRM is refused, not circumvented. Protected streams are detected and declined. This is deliberate.
• Firefox cannot stream downloads to disk (no File System Access API), so large files are buffered in memory there.
• Platform adapters break when platforms change. Reports on the issue tracker are the fix.

Privacy

No analytics, no telemetry, no error reporting, no account, no remote configuration. Settings stay in your userscript manager's local storage. Requests to platform CDNs carry the cookies your browser already has, exactly as the page's own requests do — the script neither reads nor transmits them.

Licensed GPL-3.0-or-later, so any modified version distributed to others has to ship its source under the same terms. A fork with a tracker bolted on cannot stay closed.