Choose An Integration

Graft exposes three developer surfaces. They use the same repository model and can be combined in one product.

Start with	Choose it when
Graft CLI	You are evaluating Graft, writing scripts, or can run short-lived subprocesses.
Node.js SDK	Your application needs a resident in-process repository session.
Remote service	Repositories must synchronize through infrastructure you operate.

Graft CLI

The CLI is the shortest path to a working repository. It is also the supported automation boundary for shell scripts and services that consume --json output.

Run the CLI quickstart Create a repository, stage a SQLite database, inspect a row diff, and restore it.

Node.js SDK

Use @eidos.space/graft when a Node.js application needs to keep repository state resident across calls.

Open an SDK session Install the package, initialize a repository, read status, and create a commit.

Remote service

A remote stores repository objects, refs, SQLite snapshot data, and external payloads outside the local repository. Use the HTTP packages to add Graft sync to an existing service, or connect the CLI and SDK to filesystem or S3-compatible storage.

Choose a remote backend Compare filesystem, object-storage, and HTTP remote integrations.

Then learn the model

Once the first integration works, read only the layer you need:

Architecture guide Understand repositories, snapshots, staging, merges, and recovery.

Specifications Use the normative contracts when implementing or validating an adapter.