setup-gbrain

To set up a local brain for your coding agent, you install the gbrain CLI and initialize a local memory database using either PGLite or Supabase, enabling the agent to manage memory and policy.

You configure per-remote trust policy during the gbrain setup process by registering MCP and capturing the specific trust settings, securely enabling cross-machine usage for your coding agent.

The best way to bootstrap an agent memory system from zero is running a one-command setup workflow that installs the CLI, initializes the brain database, and registers MCP so the agent can interact immediately.

Yes, you can use gbrain within your gstack workflow by running the setup-gbrain process, which integrates the local brain initialization and MCP registration directly into your automation environment.

You do not need a Supabase instance to start an agent memory brain; the setup supports PGLite for a fully local initialization, or Supabase if you prefer a managed database backend.

If your coding agent cannot call gbrain after installation, the setup likely failed to register MCP or capture the per-remote trust policy, which are required steps to securely enable agent interaction.