Back up and sync

Your files are the durable part of the brain. Keep an off-machine copy of the repository before you worry about hosting an interface.

Back up the repository

A private git remote is a simple place for a second copy. It preserves your committed knowledge and change history. Confirm that the remote is private, then make sure the files you intend to keep are actually committed and pushed.

Gitignored secrets and other untracked files are not in that backup. Large assets may need a separate storage policy. Think through what you would need to restore on a new machine, and check that those pieces have a backup too.

The search database is disposable: brain index --force rebuilds it from the files. Keeping the database is not a substitute for backing up the repository.

Use sync to reconcile changes

From the main branch of your brain, with its origin remote configured:

brain sync --human

Sync commits recognized content, pulls, merges and pushes. Some files or conflicts still need judgment. It can invoke your configured agent runner; if that runner or its account is unavailable, complete the handoff before treating the sync as finished.

For the exact merge policies and commands, use the sync reference on GitHub (external site: github.com).

Add phone access as a separate step

An optional host can provide the rich interface over your brain. That host supplies authentication, the agent backend and the PWA shell. It is separate from the private content repository and has its own configuration.

The public hosting starter is still in development. A local brain and a private remote work without it; UI packages are available for people building their own host. Cached offline recording is limited, and agent work and server history require a connection.

The hosting reference on GitHub (external site: github.com) covers the technical boundaries, and media guidance (external site: github.com) explains the tradeoffs for large files and git history.