Workspaces — Team and Personal

Everything in Portuni happens inside a workspace: one graph, one set of mirror folders, one server behind them. There are two kinds, and they are named by what they are for, not by where the database sits.

	Team workspace	Personal workspace
For	an organization: shared graph, people signed in, permissions	one person on one machine
Who reaches the graph	every teammate, through the organization’s central server	only you, through the server embedded in your desktop app
Sharing	built in: per-user permissions, per-node visibility, Drive as the shared file store	none, by design; nothing leaves the machine
Sign-in	Google account	none
Set up by	the organization (Team Setup), then each teammate joins with a server URL	you, in a few minutes (Setup)

A personal workspace is the same Portuni server in a box: what works there works the same way in a team workspace, and the desktop app, MCP tools and agents behave identically. The team workspace is the one Portuni is built around; the personal one is for trying it out and for working alone.

One desktop can hold both. Workspaces are created in Settings, each with its own sidecar, port and credentials, so a personal workspace for your own notes can sit next to the team workspace of the organization you work in.

The parts of a team workspace

A team workspace always involves two processes, and the docs keep their names apart:

• Central server — the Portuni server the organization runs (api.portuni.com or your own host). It owns the graph database, checks who you are and what you may see, holds the Drive credentials and watches the shared drive for changes. Teammates’ apps and agents talk to it; nobody holds a database token or a Drive key on their laptop.
• Sync agent — the sidecar your desktop app runs next to a team workspace. It has no graph database of its own. It keeps your local mirror folders, watches them, moves file bytes between them and the central server with a per-device token, opens files from the mirror in the editor, and runs agent tasks on your machine. Before you sign in, the sync agent is not running and the features that need it say so.
• Device-local — the things the sync agent handles itself instead of forwarding to the central server: mirror folders, file content and file actions, sync status and sync runs, the runner registry, and starting or steering a task. In a personal workspace the same things are simply handled by the embedded server.

In a personal workspace there is one process, the embedded server, and it plays every role at once.

What runs where

	Central server	Your device in a team workspace	Personal workspace
Graph (nodes, edges, events, file records)	yes	asks the central server	yes, own database
Mirror folders and the file watcher	no	yes (sync agent)	yes
File content in the editor	serves it from Drive when you have no mirror	from your mirror, else via the central server	from your mirror
Google Drive	yes, one service account	through the central server	not available
Agent tasks	records them	runs them	runs and records them
Permissions and visibility	enforced here	enforced by the central server	one user, nothing to enforce
MCP for your CLI agents	proxied from the sync agent, plus remote MCP clients with OAuth	the sync agent’s local endpoint	the embedded server’s endpoint

What a given deployment offers beyond the graph (Drive, remote MCP clients, routines, remote hosts) depends on how the organization configured its central server and on what has shipped (Project Status & Roadmap); it is not part of either kind’s definition.

The one thing people mix up

“Which kind of workspace” and “syncing files to Google Drive” are two different questions. The first is about how your app reaches the graph. The second is about where file bytes go, and only a team workspace has anywhere for them to go. The two planes are explained in Files: the two sync planes.

Words you will see in the settings and in code

In the docs	In the app and code
team workspace	data_mode: "central", “Join a team” in onboarding
personal workspace	data_mode: "local"
central server	PORTUNI_AUTH_MODE=google, the server_url in a team workspace’s settings
sync agent	PORTUNI_AGENT_MODE=1, the sidecar of a team workspace
device-local	the sync_agent_down answer (HTTP 501) when the sync agent is not running yet: sign in