Channel

MCP: your workspace, inside your AI tool.

secondbrain exposes a remote MCP server. Connecting it to Claude, to ChatGPT, to Gemini or to any client that speaks the protocol gives it access to your notes, tasks, documents, recaps and the whole Atlas — to query, to analyze and to write. This is the channel for the things that do not fit in a chat: long-range planning, cross-cutting analysis and automation.

Getting it running

What MCP is, in one line

MCP (Model Context Protocol) is the standard that lets an AI tool use external services as its own tools. Instead of copying and pasting context into a prompt, the model asks for the data it needs, when it needs it.

The practical difference is enormous: an analysis over two hundred documents does not fit in a prompt, but it does fit in twenty tool calls the model decides to make on its own.

Connecting a client

The server is remote — nothing to install — and answers at a single address:

https://api.secondbrain.media/functions/v1/mcp

There are two ways to authenticate, and you pick by client:

OAuth

The client opens the secondbrain sign-in screen, you authorize, done. This is what Claude, ChatGPT and Gemini use when adding a remote connector: no key to copy anywhere.

API token

For clients that expect a fixed credential: generate a token in Account settings → MCP, paste it into the client, revoke it whenever you want.

Which client?

On the web, on the desktop and on the phone. It goes over OAuth, so there is no token to copy.

The shortest way: secondbrain is in the Claude connectors directory, so you find it by name and there is no address to paste.

Open secondbrain in the Claude directory →

Or, if you prefer to do it by hand:

1
Open the connectorsIn Claude: Settings → Connectors, then Add custom connector.
2
Paste the addressThe name is yours to pick — "secondbrain" does fine — and the URL is the one above.
3
AuthorizeClaude opens the secondbrain sign-in: you log in with your account, approve the access and come back to Claude with the connector already connected.

From there on, the secondbrain tools are switched on per conversation from the chat tools menu.

Clients configured through a command or a file — Claude Code, Cursor, Windsurf, n8n, an agent of your own — do not do OAuth: they go with a token.

1
Generate the tokenIn the app: your name → Account settings → MCP, pick your tool and hit Generate token. It is shown once, already placed in the command or the block, with a button to copy it.
2
Paste it into the clientIn Claude Code it is a terminal command; in Cursor, a block in ~/.cursor/mcp.json. In other editors the file changes, the block does not.

claude mcp add --transport http secondbrain https://api.secondbrain.media/functions/v1/mcp \
  --header "Authorization: Bearer sbk_your_token"

{
  "mcpServers": {
    "secondbrain": {
      "url": "https://api.secondbrain.media/functions/v1/mcp",
      "headers": {
        "Authorization": "Bearer sbk_your_token"
      }
    }
  }
}

The MCP tab of account settings, with a token expanded and the Claude Code configuration block — The **MCP** tab of your account settings.

Clients it works with

Any client that supports remote MCP servers over HTTP, including an agent of your own built with the protocol SDKs. The server is published in the public MCP registry, so clients that consult it find it on their own.

What you control

An OAuth connection sees your workspaces; a token, only the workspace you generated it in. Neither reaches anything else: the same boundary the app has.
Each connection — a token or an app you authorized — is cut separately in Account settings → MCP → Your connections, without touching the others or ending your session.
Tools respect your plan: what is not included cannot be done over MCP either.
Every write is logged exactly as if you had done it in the app, with its history and its undo.
You can disconnect the server from the client at any time: nothing keeps running on our side.

Cards in the chat

What you see when the model uses secondbrain

In Claude — web, desktop and phone — and in ChatGPT, what secondbrain brings back is not just text: the chat draws a card with your day, a task, a recap or your Atlas, with the app's colors, icons and type. And you can act on the card without typing.

The model reads the same data as before; the card only shows it. Where there is no screen — voice mode, Claude Code, an agent of your own — the answer is still text.
On a computer, the cards with several parts — your day, what your memory knows, an Atlas entry, your settings — use the width in two columns. On the phone they go in one, with the secondary parts folded.
If you connected secondbrain earlier and don't see them, disconnect the connector and connect it again: the client keeps the tool list from when you connected it.

The views

There are more than thirty: almost everything the model can read or do has its own. These are the ones you'll see most.

What did you ask?

"What do I have today?": the agenda as in the app's calendar — with the now line and the button to join the call —, the day's tasks and what is overdue.

What you can do from a card

Tasks: mark them done, move them to in progress or back to to-do.
Recaps: turn each action item into a task, undo it or dismiss it.
Settings: every switch saves right away.
Open: takes you to the exact place in the app — the task, the recap, the Daily day, the Atlas entry.
Join an event's call, or open a clip's original.

What a card writes goes through the same tools the model uses, with your permissions and your plan: what the app doesn't let you do, the card doesn't either.

Day to day

The tool catalogue

More than a hundred tools, grouped by family. A connected client sees all of them and picks the one it needs.

Family	What it can do
Search	Ask the whole workspace in one go — it returns the passages, the Atlas entries and the open tasks that matter — and search by word and by meaning
Daily notes	Read a day's note, write it, list recent ones, search across all, see the provisional tasks, close and reopen the day
Tasks	List, read detail, create, update, complete, reopen, delete, comment, list the completed ones and read the task history
Boards and deliverables	List the boards and their tasks; list, create and update deliverables
Calendar and reminders	Read the agenda; create, change and delete Secondbrain events — Google, Outlook or iCloud ones are read-only —; list reminders and cancel one
Documents	List the tree, read, create, update or edit in parts, copy, move, delete, search, see and restore versions, and link or unlink entities
Recaps	List meetings and groups, read a whole one, confirm or dismiss its action items, create, update, delete, and manage meeting types
Atlas	Search entities, read an entry and its wiki, create, update, delete, merge, manage relations and tags, read definitions and an entity's tasks
Canvases	List, read, create, update and delete
Files and clips	List files, all of them or by document, rename and delete them; list and read clips
Artifacts	Save the conversation's artifact, edit it in parts, append to it, list, read, restore a version and delete
Reports	List, read the latest or any one, generate a new one, delete, and check AI usage
Workspace	List your workspaces, read and update the current one, and read its relations
Account and settings	Read and update the profile, check the plan, usage and storage, and change settings: notifications, language, theme, time zone, day closing and digest
You	Your portrait from your meetings: how much you talk, your tone, your strengths and what to improve
What the app does	For recording, dictating or connecting something that happens in the app: the steps and the right download

It also comes with recipes for the usual requests — prepare a meeting, catch up, plan the week, close the day — which the model follows when the request fits.

What it is actually used for

Quick capture is already handled by mobile and chat. MCP is for the rest: what needs to read a lot and think for a while.

Complex planning

"Look at my open deliverables, my meetings for the next two weeks and my real load, and build me a quarter plan." The model reads, cross-references and proposes; you decide what gets written.

Atlas analysis

"Which clients have stalled projects, and who on my team is connected to them?" It walks the graph, reads the entries and answers from what is there, not from an impression.

Synthesis over lots of material

Reports that cross months of recaps, notes and documents. The model asks for what it needs across several rounds instead of trying to cram everything into one prompt.

Automation

Routines that run on their own: prepare Monday's summary, open tasks from an email, archive what closed, create the entries for a new client.

Bulk loading and migration

Bring content over from another tool, create hundreds of documents with their structure, or register an entire Atlas from a list.

Writing with your context

Draft in Claude, ChatGPT or Gemini with your documents and recaps as the source, and save the result straight into your memory.

By voice, while driving

With secondbrain connected to ChatGPT, voice mode — on the car screen or on your phone — lets you talk to your memory with your hands on the wheel: what was said in a meeting, what is on your calendar, what is due. And in the same conversation you ask it to take care of it.

What do you ask?

What someone said in yesterday's meeting, and a task with a reminder so you do not forget.

A woman driving asks ChatGPT whether Diego said he would send the spec in yesterday's meeting; secondbrain reads the recap and then creates the task "Call Diego about the spec" with a reminder on Friday at 10

Artifacts

What the model builds in the conversation — a prototype, a deck, a document — is saved just by asking: "save this artifact to Secondbrain". It lands in your workspace with its origin on record — Claude or ChatGPT —, and afterwards the model can read it, edit it in parts or go back to an earlier version, from any client.

For it to also update itself every time it changes in the chat, there is the Web Clipper.

Combining it with other tools

The point of MCP is that your AI client can have several servers connected at once, and use them in the same conversation.

With secondbrain plus your calendar, your mail, your repository or your database connected, a single request can read your work context here, cross it with what is over there and leave the result written where it belongs. secondbrain contributes the part none of the others have: what you were working on, with whom and why.

Integrations

MCP

External AI clients (Claude, ChatGPT, Gemini and others) can query and operate your workspace over the MCP protocol, only with the token you generate.

Shall we try it?

Create your free account and use all of this today.

Start for free →Log in