Skip to content

Team Setup

Jump to bottom
Rahil P edited this page Aug 29, 2026 · 7 revisions

Team Setup

Second Brain can be a private store for one person or a shared brain for a small team. This page covers turning on team mode, adding people, and what changes when you do.

If you are the only person using your brain, you do not need this page. Team features stay completely out of the way until you add someone.

How team mode works

A team brain has two layers:

  • Personal — your own memories. Nobody else on the team can see them, search them, or reach them through any AI client. This is where everything lands by default.
  • Shared — the company layer. Everyone on the team can find these.

Sharing moves a memory from personal to shared. There is never a second copy, and you can move it back. A shared memory can only be edited or deleted by the person who shared it, or by an admin.

There is no flag to set and nothing to redeploy. Team mode turns on the moment your brain has more than one person in it, and the dashboard shows the team features from that point on.

Team mode is effectively one-way. Removing everyone later does not return the brain to solo mode, because removed people are kept as tombstoned records so the authorship of anything they shared survives. The team tab stays visible. Nothing breaks and no memory is lost — but treat "add the first member" as a decision you are keeping.

Before you start

You need:

  • Your Worker deployed and reachable, and your owner password (the AUTH_TOKEN you set at install).
  • A way to send each person their sign-in token privately — a password manager, a DM, anything that is not a shared channel.

Setting up a new brain as a team brain

  1. Install normally. Follow the Setup Guide — desktop app, one-click deploy, or manual. There is no separate "team" install.
  2. Open the dashboard and sign in as the owner.
  3. Name the team. Go to the Team tab and rename it from the default. The name appears in every member's app and in the activity log, so it is worth doing before anyone joins.
  4. Add your first member — see Adding a member below. Team mode turns on as soon as this succeeds.

Turning an existing brain into a team brain

Your memories are safe. The upgrade is additive: no memory is moved between layers, deleted, or made visible to anyone else.

  1. Upgrade the Worker to a version that supports teams, using How to Upgrade. Nothing else changes yet.
  2. Open the dashboard once as the owner. On the first request after the upgrade the Worker migrates its own schema — it creates the team tables, gives you an owner identity, and files every existing memory into your personal layer. Your password keeps working unchanged and you do not have to sign in again.
  3. Check that nothing changed. The dashboard should look exactly as it did before: no team tab content, no new prompts, all your memories present, and search working as usual. Until you add someone, your brain is still solo.
  4. Name the team, then add your first member below.

What the upgrade does to your existing memories

  • Every existing memory and link is filed into your personal layer. Nothing becomes shared. Nothing is deleted.
  • Your existing password becomes the owner account. No new credential to store.
  • Semantic search keeps working over everything you already had.
  • The migration runs once, automatically, on the first request. There is no command to run and no maintenance window.

Adding a member

  1. In the dashboard, open the Team tab and choose Add member.
  2. Enter their name, optionally an email, and pick a role:
    • Member — reads and writes their own personal layer, reads the shared layer, and can share their own memories into it.
    • Admin — everything a member can do, plus managing people and seeing the team's activity log.
  3. Copy the sign-in token immediately. It is shown once and stored only as a hash — nobody, including you, can retrieve it afterwards. If you lose it, use Reset token to issue a new one.
  4. Send it to them with Copy invite message, which fills in your Worker URL and instructions, or the email option.

They sign in by entering the Worker URL and that token — in the desktop app, or in the dashboard directly. They get their own personal layer immediately.

What a new member can and cannot see

A member joining an existing brain sees none of your history. They start empty.

Owner / admin Member
Their own personal memories yes yes
Your personal memories no no
The shared layer yes yes
Who is on the team yes, with details names and roles only
Other people's private memory counts admins only no
Add, suspend or remove people admins only no
The team activity log admins only no

Nobody can read anyone else's personal layer — that includes you. Being an admin lets you manage people, not read their memories.

Sharing a memory

Open any memory and choose to share it. It moves to the shared layer and everyone on the team can find it. You can move it back to personal at any time.

Only the person who shared a memory — or an admin — can edit or delete it once it is shared. Everyone else can read it and link to it.

Where new memories land

Each person has a capture default, set on the Team tab:

  • Personal — everything they capture stays private.
  • Shared — everything they capture goes to the team.
  • Auto — follows the organisation default the admin has set.

The composer shows which one applies before you capture, so "Auto" is never ambiguous.

Managing people

All of this is on the Team tab, and admins only:

  • Reset token — issues a new sign-in token and invalidates the old one. Use this if someone loses theirs or you suspect it has leaked.
  • Suspend — blocks access without removing the person. Their memories stay put. Reversible.
  • Remove — revokes access permanently and deletes that person's personal memories. Anything they shared with the team stays, still credited to them. This cannot be undone.

Removed and suspended people keep their names on anything they shared, so the team's history stays readable.

You cannot remove your own account, and you cannot remove the last active admin — the brain would have nobody who could manage it.

Roles

  • Admin — manage people, see the activity log, edit or delete anything in the shared layer, set the organisation-wide capture default.
  • Member — everything else: their own personal layer, the shared layer, and their own capture default.

The owner (whoever holds the original AUTH_TOKEN) is always an admin.

Frequently asked

Can I go back to a solo brain? Not really. Removing everyone revokes their access and deletes their personal memories, but the brain stays in team mode because removed people are kept as records so shared authorship survives. Nothing breaks; the team tab simply stays visible.

Will my colleagues see my existing memories? No. Everything you had before becomes your personal layer, and personal is private to you. Only memories you explicitly share become visible.

Does each person need a Cloudflare account? No. Only whoever deploys and updates the Worker needs Cloudflare access. Everyone else just needs the Worker URL and their token.

Can a member update the Worker? No. Updating the Worker redeploys it to Cloudflare, which only the owner's account can do. Members are told when the brain is behind rather than being offered a button that cannot work for them.

What happens if two people share the same thing? Both memories exist. Duplicate detection works within a layer, so near-identical shared memories are flagged the same way duplicates always are.

Is there a limit on team size? Nothing enforced in the product. The practical limit is your Cloudflare plan's free-tier budgets — a handful of people is comfortable.

Using the CLI and other clients with layers

The brain CLI takes a -w, --workspace flag on remember, recall and list:

brain remember "We settled on Cloudflare Workers for the API" --workspace shared
brain remember "Draft feedback for Tuesday's review" --workspace personal
brain recall "deployment decision" --workspace shared
brain list --workspace shared --limit 20

Values are personal and shared. Leave the flag off and the brain applies your own capture default, exactly as before — so nothing you already do changes. brain list marks shared rows with shared · <author>; personal rows are unmarked. A note captured offline keeps the layer you asked for and syncs into it later.

AI clients connected over MCP work the same way: the remember, recall, list_recent and share tools all accept a workspace argument, so you can tell an assistant to save something to the team layer directly. Over the REST API the same parameter is available on POST /capture and GET /list — see the API Reference.

See also

Clone this wiki locally