A page and a server, or a peer you keep.
PairDrop, the maintained fork of Snapdrop, sends files between two browsers on the same network, and across networks with a room code or paired devices, through a public server that connects them and relays the bytes when a direct link fails. KEIBIDROP keeps saved peers, so two computers reconnect on an 86 character fingerprint exchanged once, the bytes go direct or through the bridge and are never stored, and it mounts the other side's folder so you open one file. A browser peer covers the device with nothing installed.
| KEIBIDROP | PairDrop | |
|---|---|---|
| Install | A desktop app, or the browser peer | None, a browser page |
| Saved peers | Yes, a fingerprint exchanged once | Paired devices, kept in the browser |
| Across networks | Yes, direct or through the bridge | Yes, through its server |
| Works when the public server is down | An open session keeps running, and one local network pairs by multicast | No, the page needs the server |
| Browse the other side | Yes, it mounts | No |
| Open a file without receiving the folder | Yes | No |
| Account | None | None |
| Encryption | End to end, X25519 with ML-KEM-1024 | WebRTC, encrypted between browsers |
| Source code | Open, MPL-2.0 | Open, GPL-3.0 |
During pairing each side leaves an encrypted note on the relay and fetches the other side's note. No file bytes pass through it: they go direct when the two machines can reach each other, and through the bridge when they cannot. On one local network the peers find each other by a multicast announcement and the relay is not involved. A session that is already open keeps running if the relay goes down. PairDrop's public server pairs the two browsers and relays the bytes when the direct connection fails, so when it is busy or down the page has nothing to connect through. PairDrop can be self-hosted. The paths are described in Network paths.
Open web.keibidrop.com, add files, and pair with a desktop that runs KEIBIDROP. The desktop sees the files in its mounted folder and pulls only the bytes an application reads; a terminal or an agent does the same through kd. The other direction works too: the desktop's folder is listed in the browser and a file downloads when you ask for it. Two browsers can pair the same way.
KEIBIDROP mounts the other machine's folder as a folder on yours, and a read pulls only the bytes it touches, so you open the one file you came for instead of receiving a whole directory. It is a real filesystem, so Premiere, Resolve, Blender, grep and git open a path on it the way they open a local one.
PairDrop's page needs its server to pair the browsers and, when a direct connection fails, to relay. KeibiDrop peers on one network find each other by multicast, an open session keeps running if the relay is down, and a new pairing across networks waits for the relay.
On the machine that shares a folder, yes. The other side can be a browser: the web peer pairs with the desktop and moves files both ways.
Yes. KeibiDrop mounts the folder, so you browse it and pull only the files you open.
Compared with LocalSend · Compared with croc · Network paths · Install