Local and Cloud Brains

MindMux uses one knowledge model in two storage contexts. A local Brain lives with a project on your machine. A Cloud Brain lives inside a Cloud Project and can be reached through the Cloud app or a compatible MCP client.

What stays the same

Both forms organize knowledge into project-level root pages and structured Pages. Pages carry metadata, a current compiled_truth, and an append-only timeline. This means the mental model travels with the knowledge even when the place you work changes.

Local Brain

Choose a local Brain when filesystem ownership, offline access, and ordinary Git workflows are most important. Its Markdown files sit under the project root, remain inspectable without MindMux, and are a natural fit for the desktop app.

Local operational state, such as session logs and task runtime data, remains separate from the Brain.

Cloud Brain

Choose a Cloud Brain when a shared, browser-accessible source of knowledge is more important. It belongs to a Cloud Project in a Workspace, supports member and Project-level access control, keeps server-side revisions, and can be selected explicitly by an MCP client through OAuth.

Cloud does not turn a Brain into an unstructured file bucket. The same distinction between current understanding and historical evidence still applies.

The relationship is explicit

Cloud does not automatically discover a local folder or decide which Brain an agent should use. When connecting an MCP client, select or create the target Cloud Brain. When using the desktop app, connect or migrate through its Cloud settings so the user can review the target and the result.

Avoid treating a local and Cloud Brain as two independent editable copies of the same project. Decide which one is the source for a workflow, then use the supported connection or migration path to change that decision deliberately.

A practical choice

The choice is operational, not philosophical: both are designed to preserve structured, durable project knowledge.