Your memory.
Your repository.

A brain you own

Serenity stores personal memory in a Git repository. Sources and memory records are inspectable, and Git preserves a history of committed changes. The runtime search index supports retrieval; the repository is the durable record.

Choose a backup location

Initialization installs a post-commit push hook. If you add a private remote and configure its upstream, commits can be pushed there. Without a remote or another backup, the local repository is your only copy. Keep personal brains private.

Understand model traffic

Full-text search works without model calls. When you enable extraction or answer composition, source or query context is sent to the provider you configure. Review that provider’s retention and data-use terms. Local endpoint configuration is available for supported adapter paths.

Credentials have their own home

Serenity’s authentication token and Gmail app password use the OS keychain. Model-provider credentials are read from the process environment. Do not write API keys into source notes or commit them to your brain.

Think beyond one file

Deleting a working-tree file is not the same as removing it from Git history, backups, or a provider’s records. If your goal is complete removal of sensitive material, account for every copy. The connector and source lifecycle documentation describes supported behavior; do not assume automatic erasure across systems.

This website is separate from your brain

The adoption chat answers questions about Serenity’s public documentation. It does not connect to your personal brain. Do not paste private notes, credentials, or confidential source material into it.