A lot of "video downloader" tools work the same way: you paste a link, it goes to their server, their server fetches the file and hands it back to you. That's an extra stop your data makes on the way — someone else's server now knows what you're downloading. Doomly doesn't have a server in this path at all, which is less a privacy feature and more a direct consequence of how the extension is built: it has no backend, period. Settings, themes, filters — all of it lives in your browser's own storage, never sent anywhere.
The easy case: photos and short videos
When you download a photo or a short video, your own browser fetches the file directly from X's media servers — the exact same servers your browser already talked to in order to show you the media in the first place. Nobody in between sees the request. It's the same trust boundary you already crossed by opening the post.
The hard case: long video
Longer videos on X aren't a single file. They're HLS — chopped into small segments listed in a playlist, and often the audio is a completely separate track from the video. A downloader that doesn't handle this either fails outright on long videos or hands you a file with no sound, which is worse than admitting it can't do it.
Doomly fetches every segment of the video track and, if one exists, every segment of the separate audio track, then combines them into a single .mp4 — entirely inside your browser tab, using a small library built for exactly this. It's packet passthrough, not re-encoding: the actual video and audio data isn't being decoded and redrawn, just repackaged into one container. That keeps quality identical to the source and makes the whole thing fast enough to not feel like a "processing" step at all.
If the mux step fails for any reason, you still get the video — just without audio, and a warning telling you so. Doomly would rather hand you an imperfect file than no file.
Why this shapes what downloads can promise
Because there's no server relaying the request, there's also nothing in this path that can go down on its own, get rate-limited separately from X, or get blocked by a service outage that has nothing to do with your connection. Downloads can still fail for reasons specific to a given post — long videos with an unusual segment layout are the main one we're still tightening up — but whatever goes wrong, it's a bug in how Doomly reads that specific video, not a second service with its own uptime to worry about.
Why it's a paid feature anyway
If there's no server cost, why charge for it? Because the cost isn't hosting — it's the engineering behind getting HLS segments, separate audio tracks, and muxing right without silently producing broken files. That work is what Pro and Shield subscribers are funding, the same way they fund everything else that isn't free-tier scroll mode.