# Build your second brain for Claude

**The CV Method · The Playbook · v1 · Oct 2026**

*by Corina · @the_cv_method · hello.thecvmethod.com · current as of Oct 2026. The AI space moves fast, so double-check tool names.*

Stop re-explaining yourself every chat. Put who you are, how you work and what you're working on into a folder of plain notes that Claude reads first, then skip straight to the task. This is how I set mine up in Obsidian, plus the copy-paste files to start yours today.

## What you're building

A "second brain" sounds fancy. It isn't. It's **a folder of short notes**: who you are, how you write, your projects, your rules, what's decided and what's still open. Claude reads the right notes **before** it starts, so you stop typing the same setup into every chat. No magic and no special AI database. It's just **context Claude reads first.**

What it saves: the long setup prompt at the start of every chat (where your files live, the full backstory on a project, how you like things written), the room that setup eats in Claude's context window, and most of all **your time.**

## Rule zero

If you've explained it to AI twice, it belongs in the vault.

Every time you catch yourself re-typing context, that's a note you haven't written yet.

## Step 1 · Pick one home

### 01 · Make one folder that holds everything

I use **Obsidian**: a free note app that stores everything as plain Markdown files in a folder on your computer (it calls the folder a "vault"). You get links between notes and a graph view, and the files stay yours. Claude can read them directly because they're just text.

Obsidian is optional. Any folder of **.md** files works the same way. Start simple and get fancy later.

> Brand new to AI? Start with a Claude Project instead (Step 2 shows how), and move to a folder once your notes outgrow it.

## Step 2 · Write the front door

### 02 · A short CLAUDE.md at the top of the folder

This is the front door: who you are, the house rules, and where to find things. In **Claude Code** (and the Claude desktop app working in a folder), a file named **CLAUDE.md** in the folder you open is loaded automatically at the start of every session. You don't have to point to it.

On **Claude.ai**, paste the same text into a Project's **instructions** and upload your notes as **project knowledge**. Projects are on every plan, including free.

> Keep it short. Anthropic's docs suggest under about 200 lines per CLAUDE.md, because everything in it loads every time and longer files are followed less consistently.

## Step 3 · Add the core notes

### 03 · About me · how I write · projects · decisions · open items

**about-me**: your role, your business, who you serve. **how-i-write**: your voice, the words you never use, a good example. **projects/**: one short note per project (goal, status, where the files live, who's involved). **decisions**: what's been decided and why, so Claude stops re-asking. **open-items**: what's still in flight.

> Steal from your own chats. The messages where you corrected Claude are your best notes, already written.

## Step 4 · Route it

### 04 · A "for this, read that" map in the front door

Don't make Claude read everything every time. Add a short **routing table** to CLAUDE.md: "writing a post → read how-i-write", "working on Project X → read projects/project-x". Claude opens only the notes the task needs, and that keeps your context window free for the actual work.

> This routing is what lets me skip prompt and context engineering and go straight to the task.

## Step 5 · Write back before you stop

### 05 · End each session by updating the notes

Before you close a chat, have Claude add what changed: new decisions, things finished, things still open. That's what makes it a **brain** and not a filing cabinet. The next session picks up today instead of starting from a blank slate.

> Read what it wrote before you keep it. You approve what goes into your brain.

## Step 6 · Keep it lean

### 06 · Prune once a month

Delete what's stale, merge what overlaps, and fix anything that contradicts something else. If two notes disagree, Claude may follow either one. A small, current brain beats a big, messy one.

## The starter folder · copy this

```
my-brain/
├── CLAUDE.md            ← the front door (Step 2)
├── about-me.md          ← who you are, who you serve
├── how-i-write.md       ← voice, nevers, one good example
├── decisions.md         ← what's decided + why
├── open-items.md        ← what's still in flight
└── projects/
    ├── project-a.md     ← goal · status · where files live
    └── project-b.md
```

## The prompts · copy these

### A · your CLAUDE.md (fill in the blanks)

```
# CLAUDE.md: read this first

## Who I am
<one or two lines: role, business, who you serve>

## House rules
- <how you want answers: short, plain English, bullets...>
- never <invent a number / send anything without my OK>
- ask before <anything you want approval on>

## Where things live (read only what the task needs)
| If the task is...         | Read first                  |
|---------------------------|-----------------------------|
| writing anything          | how-i-write.md              |
| <project A>               | projects/project-a.md       |
| planning / priorities     | open-items.md, decisions.md |

## Before you stop
Update decisions.md and open-items.md with what changed. Show me first.
```

### B · have Claude set it up for you (paste this)

```
Help me build a second brain you can read before every task.

Interview me one question at a time about: my work, who I serve,
how I like to write, my current projects, decisions already made,
and what's still open. Then write these as short Markdown files:
CLAUDE.md (with a "for this task, read this file" routing table),
about-me.md, how-i-write.md, decisions.md, open-items.md, and one
file per project in projects/. Keep each file short and specific.
Don't invent anything I didn't tell you.
```

### C · the write-back (paste at the end of a session)

```
Before we stop: list what we decided, what got done, and what's
still open in this session. Then show me the exact edits you'd
make to decisions.md and open-items.md. Wait for my OK.
```

## Where it works

| Where you use Claude | How it reads your brain |
|---|---|
| Claude Code | Open the folder. CLAUDE.md loads automatically at launch, and Claude opens the other notes when the task calls for them. |
| Claude desktop app | Give Claude access to the folder, and it reads the notes from there. |
| Claude.ai Projects | CLAUDE.md text goes in the project instructions, and the notes go in as project knowledge. Re-upload a note when it changes. |

Features change. Check code.claude.com/docs (memory) and support.claude.com (Projects) for the current details.

## Keep it honest & safe

- **No passwords**, keys or account numbers in any note
- Client details **anonymized**
- **You approve** every write-back
- **Short notes** beat long ones
- Claude **proposes**, you decide

## The starter checklist

- [ ] **One folder** (or one Claude Project) picked
- [ ] A short **CLAUDE.md** front door
- [ ] The **five core notes** started
- [ ] A **routing table**: for this task, read that
- [ ] The **write-back** at the end of each session
- [ ] A monthly **prune** on the calendar

## Want your second brain set up for you?

I build the front door, the routing and the core notes around how you actually work, so Claude starts every task already knowing your business. That's the CV Method.

**hello.thecvmethod.com/start**