Powered by

Quickstart

This is a ~60-second round-trip with no cloud account and no setup: snapshot a directory, push it to a local store, pull it back somewhere else, and confirm it is byte-for-byte identical. The only thing you need is the snapdir binary — see Install if you don't have it yet.

A snapshot is identified by the BLAKE3 hash of its manifest, so identical content always produces the same snapshot ID on any machine. Every object is re-hashed and verified on fetch, so a successful pull is a proof of integrity.

1. Create a directory and read its ID

Make a directory with some content, then ask snapdir for its content-addressed ID. Nothing is stored yet — snapdir id just hashes the directory:

mkdir -p demo && echo hello > demo/a.txt

snapdir id ./demo

The printed ID depends only on the content. Run snapdir id ./demo again (or on another machine with the same files) and you get the exact same ID.

2. Push to a local store

A store is addressed by a URI of the form protocol://location/path. For a local round-trip the protocol is file://, which is just a directory on disk. push uploads the snapshot and prints its ID, so capture it in a variable:

id=$(snapdir push --store "file://$PWD/store" ./demo)

echo "$id"

That is the same ID snapdir id printed in step 1. Objects are stored at content-addressed keys, so unchanged files are never uploaded twice.

3. Pull it back and verify

Pull the snapshot into a fresh directory. pull fetches every object from the store, re-hashes it, and checks it out — failing loudly if anything does not match:

snapdir pull --store "file://$PWD/store" --id "$id" ./restored

snapdir id ./restored   # prints the same $id — byte-for-byte identical

If snapdir id ./restored prints the same ID, the restore is verified identical to the original. You can also explicitly check a staged snapshot with snapdir verify.

Going further

Swap file://$PWD/store for an s3://…, gs://…, or b2://… URI and the exact same push/pull commands publish to and restore from the cloud — no other changes.