setup

To set up GBrain, run the init command with your preferred database flag, followed by the doctor and scaffold commands. This wizard-based process auto-provisions a database, injects the brain-first lookup protocol, and configures live sync.

Agent injection automatically inserts the brain-first lookup protocol into your project's AGENTS.md file. This ensures that agents reference the brain index first, maintaining data integrity during live sync with repository changes.

You do not need an external account. GBrain can auto-provision a local PGLite database. Alternatively, you can initialize the system by providing a specific connection string for an existing Postgres or Supabase database.

You need the bun runtime installed in your local environment to execute the setup scripts. Postgres or Supabase accounts are optional unless you want to connect to an external database instead of using auto-provisioned PGLite.

Run the gbrain doctor command with the json flag after initialization. This verifies your configuration and confirms that live sync is properly configured to keep the brain index up-to-date with repository changes.

Yes, you can initialize GBrain without embeddings by passing the no-embedding flag. This allows you to scaffold the system and configure the brain-first lookup protocol before integrating or requiring an embedding model.