Build your Second Brain step by step
The video above shows the whole build live, click by click; this guide is the written companion so you can move at your own pace with every command and prompt at hand.
What you'll get: a second brain working end to end — Claude Desktop, Obsidian and Claude Code installed, the Claudian plugin connecting them, your personal CLAUDE.md written from a 6-round interview, and your first project created with the new-project skill.
Stage 1 — Install
Download the two apps
Claude Desktop (from claude.com) and Obsidian. They are your brain's two windows: one thinks, the other shows the files.
Get the template folder
Download the creator's template folder zip (free in his community) — or build the structure yourself. Rename the folder to your brain's name.
Install Claude Code in the terminal
Copy the install command for your operating system and paste it into the terminal. When it finishes, verify with the command below.
Open your brain as a vault
In Obsidian: Open folder as vault → pick your brain folder → Trust.
To verify Claude Code is installed, run this in the terminal:
claude
Stage 2 — Connect Claudian (Claude inside Obsidian)
Claudian is the plugin that puts Claude into Obsidian's right sidebar.
Security habit, always: before installing any community plugin or repo, have Claude review it. A repo that's clean today can be compromised tomorrow — the one-minute review is free and stays with you forever.
Copy the plugin's GitHub link and, in the terminal, ask Claude — and tell it which vault to install into:
I want to install this Obsidian plugin: [github.com/plugin-repo]. Check this repo and make sure it's clean of anything malicious; if it is clean, install it into my vault at [~/my-brain].
Enable the plugin
When Claude finishes, restart Obsidian and, if needed, enable Claudian under Community plugins.
Configure Claudian
In the plugin settings: pick the model (Opus if you have it; Sonnet on the $20 plan), effort medium, and the permissions mode (see the note below).
Safe mode vs. yolo mode: in safe mode, Claude asks for permission before every action; in yolo mode, it bypasses the questions. Start in safe. Trust is earned by watching what it does — yolo comes later.
Stage 3 — Personalize the brain
With everything connected, ask Claude:
run the brain setup skill in the 05 skills folder
The interview
The skill interviews you in ~6 rounds (detailed below) and writes your Layer-1 CLAUDE.md from your answers.
Your personal CLAUDE.md
The file that tells Claude who you are, what you're trying to do, what its rules are, and what your goals are. It's the difference between a generic chatbot and a partner that knows you.
| Round | What it asks you |
|---|---|
| 1 | Who you are |
| 2 | What Claude should do (thinking partner) |
| 3 | Boundaries and communication style, plus rules like "tag files you create with (C)" and "never edit my notes without asking" |
| 4 | Your strengths and weaknesses (and you can ask it to "poke holes in my thinking") |
| 5 | Your goals and current progress |
| 6 | Your current projects (having none yet is fine) |
You write your goals yourself. Don't let the robot decide your destiny: AI helps you achieve the vision, it doesn't choose it for you. Arrive at round 5 with your goals already thought through.
Stage 4 — Your first project
Create it with the new-project skill
The skill asks the project name, what it is, what "shipped" looks like, who's involved, and the process — then builds the whole folder before your eyes: inputs, process, outputs, system, 05-skills, and its own CLAUDE.md.
Work inside the project
Manage vaults → Open folder as vault → your-brain/projects/<one>. Adjust the project template once (colors, plugins) and every future project inherits it.
The video's example is a finances project. This is how you answer the new-project interview, with your own details:
Run the new-project skill. Project name: "finances". What it is: tracking my personal money. What "shipped" looks like: a monthly report plus a macro analysis, generated from the bank-statement PDFs I drop into the inputs folder, checked against my pre-agreed budget of [$3,500/month]. Who's involved: [just me]. Process: I drop the PDFs in, Claude processes them and writes the report to outputs.
Phone access (Dispatch)
In Claude Desktop: Cowork → Dispatch tab. Turn keep-awake on and give it the second-brain folder as its permanent workspace. From your phone you can then, for example, save a video idea into the exact folder of a project, formatted per your rules, while walking down the street.
Dispatch is glitchy at first: get everything working on the computer before relying on the phone.
Sync with your team (Obsidian Sync)
It's a paid add-on (~$4/mo). Under Settings → Sync (core plugin): choose or create a remote vault, share it with your teammates' emails, and have them connect from their Obsidian. The green check means fully synced; phones can join too.
If you share code or HTML with the team, enable "sync all file types" — and remember you can exclude folders you don't want synced.
10-minute practice: complete Stage 1 today: download Claude Desktop and Obsidian, get (or create) your brain folder, rename it, and open it as a vault with Trust. Tomorrow, with the video beside you, connect Claudian and do the interview.