The virtual folder is a real filesystem backed by another machine. Most things that work on a local folder work on it too. Here is what that is good for.
Your repo is on your desktop and you are on your laptop. Mount the desktop's folder and use git directly:
cd ~/KeibiDrop/Mount/projects/my-repo git status git checkout -b feature git commit -am "wip" git push
Clone, status, fetch, checkout, and git-lfs work on the mount. Nothing is copied to your laptop except the bytes git reads. Good for driving a build machine from a thin laptop, or reviewing a teammate's working tree without them pushing first.
More: Git clone from another computer
Run a database off the shared folder on one machine. Stop it, remove its lock file, and start the same database off the same folder on the other machine. Every row is there.
# Machine A pg_ctl -D ~/KeibiDrop/Mount/pgdata stop rm ~/KeibiDrop/Mount/pgdata/postmaster.pid # Machine B, same data directory pg_ctl -D ~/KeibiDrop/Mount/pgdata start psql -c "select count(*) from t;"
The data directory streams on demand and the database reads the pages it needs. This is the demanding case, with fsync, random page reads, and many open files at once, and it round-trips without corruption. Good for moving a dev database between a desktop and a laptop instead of dumping and restoring.
A 10 GB video, a multi-gigabyte dataset, a disk image. Open it from the mount and it starts right away, because KEIBIDROP only pulls the parts you touch.
Scrub a video in QuickTime or VLC and it seeks and plays from the middle without fetching the whole file. Read a slice of a large CSV. head, tail, or dd skip= a big file and only that region comes over.
More: Stream files without downloading them first
Both of you mount the same shared folder. You edit a file and they see the new content on their next read. They drop a file in and it shows up on your side. Good for pairing on a project, handing off assets, or a shared scratch space, without a cloud drive syncing everything in both directions.
More: Work on files without copying them
It is not only for two people. Mount your home desktop's folder from your work laptop, grab the file you forgot, or keep one project folder that you open from wherever you are.
The mount is just a path, so any tool takes it:
ffmpeg -i ~/KeibiDrop/Mount/raw/clip.mov out.mp4 grep -r "TODO" ~/KeibiDrop/Mount/projects/ duckdb -c "select * from '~/KeibiDrop/Mount/data/events.parquet' limit 10"
Only the bytes your program reads get transferred.
More: Run ffmpeg on remote files
If you do not need a folder mounted, use Direct Transfer. Drag in a single large file, a whole folder, or a batch of files, and your peer pulls what they want. It runs at up to ~660 MB/s, needs nothing installed, and leaves nothing behind. There is no size limit and no upload step.
More: Send large files without uploading them
The kd agent CLI is scriptable and prints JSON, so cron jobs, scripts, and AI agents can drive it:
kd start kd register <peer-fingerprint> kd connect kd add ./nightly-build.tar.zst kd pull report.pdf ~/Downloads/
More: Agent CLI guide
The files stay where they are, and you work with them where you are. No syncing everything everywhere, no uploads, and no copies you have to remember to clean up. Just a folder that happens to live on another machine.
New here? Start with Getting Started.