your second brAIn

Part 1 — What a second brain is

A second brain is a folder with all of your digital information organized in text files, plus an instruction that tells the AI how to navigate and understand these files. With this second brain you can connect it to everything you use in your life (email, meetings, calendar and the documents that matter) and what you get is a virtual assistant that already knows you even more than you do.

Many people say their AI knows them because they wrote general instructions in their favorite AI’s settings. This takes that to another level: an AI that stores all the useful information about your life and is proactive about helping you do your work in a better way.

The rest of this part is what a second brain is made of and the rules that keep it healthy. Nothing here asks you to do anything yet — the installation is Part 2.

The six parts of a second brain

1. A folder of plain text you own

The only thing you need are text files, and most of them are going to be written by the AI, so don’t worry about it.

That folder will not only have facts about you. It will also have your thoughts, your experiences, the decisions you made and why, the way you like to work. If you already have a document that explains who you are (a profile you wrote for an AI, a long note about yourself) it is going to be useful. If you don’t have it, the AI will build it with you.

Because the second brain is plain text sitting on your computer, if the AI you use today disappeared tomorrow you could hand your entire context to another one with no effort. Your brain is not tied to any company, although you still have to act responsibly with information that belongs to the companies you work with. And because the files stay on your machine, your data is more private.

Without it: the AI’s memory belongs to whoever hosts it, and dies with the subscription.

2. The standing instruction

The most important file you are going to create is called CLAUDE.md. Every time you start a new chat, the AI reads this file. It holds the basics: who you are, what you’re responsible for, who matters, how you work, how to talk to you, what you already decided, where new things go. Leave out anything that changes often (the status of something urgent, for example), because then you’d be the one maintaining it.

Your main CLAUDE.md should be complete and concise: every sentence in it earns its place, and nothing that matters is missing.

There are two kinds of this file, and you should have only one. The first lives in your AI’s own settings and applies to every conversation everywhere. The second lives inside the folder of your second brain, and that is the one we will use. If you already have one in your settings, hand it to the AI during setup — it will use it to write a better one, and then you delete the old one. Two instruction files that drift apart will eventually contradict each other, and the AI won’t know which one is true.

Without it: you re-explain yourself every conversation, forever.

3. A place for everything, and you never have to use it

Your life has a shape: the things you’re responsible for, the people around them, the work you’re doing this year. The second brain mirrors that shape in folders named in your own words, and every kind of thing has exactly one home — a meeting note here, a person there, something the AI wrote for you over there.

You should know roughly what that shape looks like. That is all you need it for: to ask “what do we have on this client?” and know the answer exists somewhere. You never file anything yourself. You don’t drag documents into folders and you don’t decide where a note belongs. When you want the brain to know about a document you hand it to the AI in the conversation, or you point the AI at a folder you already keep, and the filing happens on the other side.

The structure is yours. During setup the AI proposes one based on how your life is actually shaped, and you rename whatever doesn’t sound like you.

Without it: everything the AI writes lands in one pile, and a pile is not a brain.

4. Capture in the moment

When you say something important in a chat — a decision, a fact about a person, a preference, a thought worth keeping — the AI tells you on the spot and asks for your confirmation before saving it. You never have to remember to save anything, and you don’t have to sit down to “process your notes.”

The same happens with what comes in through your email and your calendar: every day, what matters gets written into your folders (a line on the person it involves, a line in the log of the part of your life it belongs to) without you doing anything. In a month, each of your folders reads like a diary someone kept for you.

Without it: the brain depends on your discipline and your time to keep updating the AI.

5. Live connectors

For the sake of simplicity we are only going to connect email and calendar. The point is that the brain reads your actual life instead of only what you type into it. You can connect other sources later: tasks, meeting recordings, messages.

Without it: the AI only has the static information about you, not the email that arrived this morning.

6. Jobs that push

Every morning the AI welcomes you with a daily brief: what your day looks like, what you promised, who is waiting on you, what you should read. This is the part that changes how the brain feels — from a place you go to look things up, to an assistant that shows up with things you needed and didn’t know to ask for.

Without it: you are asking the AI the same things over and over.

The six laws

These are the rules that keep a second brain healthy. Every one of them was learned by Pablo through practice, trial and error.

1. The original never moves

The brain holds a note about your document, not the document. The note records what it is, why it matters, who it involves, and the full path to where the original lives. Your files stay exactly where you put them.

This is what makes the brain safe to say yes to. Nothing gets moved, renamed or deleted.

2. The brain is text

PDFs, Word documents and presentations become notes. A note about a document is smaller, searchable, and readable by any AI.

Spreadsheets are the one exception. A summary of a table loses the numbers, and the numbers are the point. Spreadsheets stay as spreadsheets, and the brain keeps a note about each one so it knows what’s inside and when to open it.

3. One home per thing, decided before the thing exists

The routing is written during setup, not improvised at save time. When something new arrives, the AI already knows where it goes, and you are never the one making the call.

4. Filing is the AI’s job

You talk; the machine files. Any design that requires you to be organized has already failed, because the people who most need a second brain are precisely the ones with no time to maintain one.

5. It has to give back more time than it takes

Any feature that adds a daily chore gets cut, no matter how clever. The test for every new piece is simple: does this save more time than it costs?

6. Every file has one author

There are files you write and files the AI writes. Your instruction file is yours. The daily brief, the log in each of your folders, the note about a document — those are the AI’s.

The rule: the AI never rewrites what you wrote, and you never edit what the AI generated. If the AI’s note is wrong, you tell it and it fixes the note. If the AI thinks your instruction file should change, it proposes the line and you approve it.

In the few files you both touch, the AI only adds lines and you only edit or remove yours. Nobody touches the other’s line.

Why this matters: when two hands edit the same file, the AI overwrites your correction or you overwrite something it needed, and from that moment the brain holds two versions of the truth. One author per file is what keeps it honest.

Part 2 — How to install it

What you need

Mac works today. Windows is written but not tested yet.

Before you start — five minutes, once

If you have never used Claude, do this first. Nothing here is technical.

1. Get the app. Go to claude.ai/download, download the Claude desktop app for your computer and open it.
2. Sign in and pick a plan. Sign in with your email. You need a paid plan that includes Claude Code: Pro is enough to start. You can upgrade from the app under Settings → Billing.
3. Make the folder. On your Desktop, create a new empty folder and call it something like My Brain. Your notes will live there.
4. Connect your email and calendar. In the app, open Settings → Connectors and connect your Gmail and Google Calendar (or Outlook if you use Microsoft). Accept read-only access. If you skip this, /setup will walk you through it.
5. Open the folder in Claude Code. In the app, click the Code tab, then choose your My Brain folder as the project. You’ll see a chat box. That’s where you type everything below.

Install

Open the Claude desktop app and open the empty folder you chose as the project. Then type these two lines in the chat, one at a time.

First, add the marketplace where the plugin lives — this is the GitHub repository, so the app downloads it from there:

/plugin marketplace add pablogzzr/your-second-brain

Then install the plugin itself:

/plugin install your-second-brain@your-second-brain

That’s the whole installation. If the app shows the commands with a prefix (/your-second-brain:setup), that spelling works too.

To update later, run /plugin marketplace update your-second-brain and the app pulls whatever is new from GitHub.

What the AI can and can’t see

In this version the AI reads three things: your email, your calendar, and whatever you bring to it — a document you drag into the conversation, a folder you point it at.

It does not scan your computer looking for new files, for two reasons. Giving an AI standing access to everything on your disk is a privacy decision you shouldn’t have to make on day one. And you shouldn’t have to save every document twice, once for you and once for the AI. When you want the brain to know about a document, you show it. That’s the only door for now.

/setup — your first fifteen minutes

Type /setup in the chat and the conversation starts. Here is what happens in it, in order.

It reads before it asks. The AI goes through your email from the last couple of months and your calendar, and comes back with what it found: the people who appear most, the projects it can see, the shape of your week. If you have an existing instruction file or a document about yourself, this is the moment to hand it over.

It asks you one question about your organization. Once, in this shape:

If you had to explain how you organize your life to someone, how would you do it? What is the actual shape of what you’re responsible for — what are the big buckets?

There is no right number. Most people land between three and seven. The AI proposes two or three options based on what it already saw in your email and calendar (for example, professional and personal, and then inside professional the two or three things you actually run). You pick one, rename what you want, and move on. Underneath, the AI keeps its own map from your names to the kinds of thing it files. You never see that map, and you never use it.

Then it asks what your data couldn’t tell it. There is no script. The AI reads first, finds the gaps, and asks only what it couldn’t learn on its own, prioritized so that a handful of answers cover most of what matters. The questions that tend to matter most are the ones no file can answer:

• What are you responsible for that nobody else can do?
• What has to be true ninety days from now for this to have been a good quarter?
• Who are you slowest to get back to, and it costs you?
• When you’re not sure about something, who or what do you check?
• When an assistant is annoying you, what is it doing?

Then it writes your instruction file. There is no fixed template, because no two people need the same one. What goes in depends on you. These are the things it can hold, and the AI picks the ones that matter for you and leaves out the rest:

• Who you are — your role, what you’re responsible for that nobody else is, what you’re building toward.
• What you’re working on — the live projects, in your words, with the people attached.
• The people who matter — the names that came out of your email, with one line each on the relationship.
• How you usually work — your patterns, including the ones that cost you. “I get excited about an idea and commit before I’ve looked at the real alternatives.” “I say yes in the meeting and regret it by Thursday.” This is the section most people skip, and the one that makes the AI useful instead of agreeable.
• How you decide — who you check with, what you trust, what you refuse to do.
• How to talk to you — length, bluntness, language, what annoys you in an assistant.
• What phase you’re in — what you’re optimizing for this year, and what advice doesn’t apply to you right now.
• Where things go — your structure and the routing rule, in one short table.
• What you’ve decided and what you think — empty on day one. It lives in its own note next to the instruction file (My decisions and thoughts.md), because the AI writes it and the instruction file is yours (see law 6). It grows every time you say something worth keeping, and it’s the reason the brain gets better instead of just bigger.

You read the draft and correct whatever doesn’t sound like you before it’s saved.

And it gives you something back before you go. With what it now knows, the AI closes with something it noticed that you should act on: a suggestion, a reminder of something important, an improvement to how you’re running your week. That is the moment it stops being a folder.

All of it, minute by minute:

Your next 45 minutes

Setup leaves you with a brain that knows you. These two commands are what fill it and what make you open it tomorrow. Both are explained in Part 3.

Part 3 — How to get the most out of it

/organize — whenever you want

Point it at a folder. The AI reads what’s there and writes a short note into your brain for each thing that matters: what it is, why it matters, who it involves, where the original is.

It never moves anything. It never deletes anything. It reads, and it writes notes into your brain. Your files are exactly where they were when it finishes. It runs in batches, tells you what it did and what it skipped, and you can stop it any time.

/today — every day

One page, one link you bookmark: what’s exceptional today, who is waiting on an answer from you, what you promised and haven’t done, and what you need to have in your head before the meeting that matters. Everything routine is left out — a brief that lists your recurring nine o’clock is a brief you stop opening. Tick what’s done and it doesn’t come back tomorrow.

At the top of the page, three numbers: how many days in a row you’ve opened it, how many things the brain filed for you this week without you lifting a finger, and how many decisions and thoughts of yours it has kept. On day fourteen, and only that day, the page says so.

While it reads your day, it writes into your brain: a line on each person who came up, a line in the log of each folder something touched. That is how the folders you named on day one fill themselves. If a document arrived by email — a contract, a deck — the page names it and offers, once, to write a note about it.

It ends with one question, in the chat: anything on your mind? Whatever you answer — a decision, a worry, an idea, something you noticed — goes into your brain. This is how your thinking gets captured without you ever opening a file.

The fourth thing, which is nothing

In any conversation inside your brain’s folder, when you say something durable — a decision, a fact about a person, a preference — the AI files it and confirms in one line. You don’t have to do anything for this one.

How you know it’s working

• Day 1 — the AI told you something useful about your own week that you hadn’t asked for. Your brain has your profile and your first notes.
• Day 7 — you’ve typed /today at least three times, and the “filed for you” number on the page is above zero.
• Day 30 — the note on what you’ve decided and what you think has more than ten lines, and each of your folders has a log you can read as a month of your life. That growth is the only real proof the brain is alive.

Day 14 is the day that matters. Installing proves nothing. Whether you still open it on day 14 is the whole test, and the page tells you when you get there.

What comes next

This plugin does one thing: it walks you through this document and builds your brain with you.

What it builds is the foundation for the full assistant — the one where the jobs run without you. A page that’s ready before you wake up, without typing anything. The loose threads found for you: who owes you an answer, what you promised and forgot. A coach that reads your meetings and your emails and tells you what you could have said better. A digest of everything worth reading from the noise you receive. A weekly close that asks you the questions nobody else will.

That is the follow-up that Pablo does.