How Do You Build a Second Brain with Claude Code and Obsidian?
Blog Post

How Do You Build a Second Brain with Claude Code and Obsidian?

Jake McCluskeyUpdated
Back to blog

The Problem Every Claude Code User Hits

You spend two hours working with Claude on a complex project. You figure out the perfect architecture. You nail the right approach for handling edge cases. You make a dozen small decisions that all fit together perfectly.

Then you close the window.

Next session? Claude has no idea any of that happened. You're starting from zero. Again.

Architecture decisions, breakthroughs, patterns that took hours to figure out. All gone when the window closes. This isn't a Claude problem. It's a memory problem. And the fix isn't waiting for a smarter model. It's giving Claude a memory system.

Step 1: CLAUDE.md, Your AI's Permanent Memory File

Create a file called CLAUDE.md in your home folder. Claude reads it automatically every single session. No configuration needed.

Write your name. Your project details. Your preferences. Anything you don't want to explain again. Think of it as onboarding a new team member who has perfect recall.

Most people treat CLAUDE.md like a settings file. They throw in a few bullet points and move on. The builders getting 10x output from Claude treat it differently. They write it like they're briefing a new hire who will never forget a single thing you tell them.

Include your coding style preferences. Your business context. The decisions you've already made and why. The more Claude knows about you upfront, the less time you waste re-explaining things.

Step 2: Obsidian + MCP, Real-Time Knowledge Access

Obsidian is a free note-taking app that stores everything locally on your computer. No cloud dependency, no subscription, no data leaving your machine.

MCP (Model Context Protocol) is the bridge that lets Claude search your Obsidian notes in real time while you work. Two servers make it happen:

  • smart-connections handles semantic search. Claude doesn't just match keywords. It understands what you're looking for conceptually.
  • qmd handles structured queries. When Claude needs specific data from a specific note, this is what finds it.

Once connected, Claude pulls up relevant notes on its own. You don't paste anything in. You don't point it at files. Claude searches your knowledge base automatically as it works.

Here's a pro tip that makes a huge difference: name your notes as claims, not categories. Don't create a note called "memory-systems.md" and dump everything in it. Instead, create notes like "memory graphs beat giant memory files.md" or "semantic search outperforms keyword matching for code context.md"

The title alone tells Claude whether a note is relevant before it reads a single word inside.

Step 3: brain-ingest, Automatic Knowledge Capture

Most knowledge doesn't start as text. It starts as a YouTube video you watched. A voice memo you recorded on your commute. A meeting where someone explained something important.

brain-ingest takes any of these sources, transcribes them locally (nothing leaves your machine), extracts the key ideas, and drops a structured note into your Obsidian vault automatically.

From a single 90-minute conference talk, brain-ingest can extract 12 to 18 key ideas and 3 to 5 frameworks. All of them findable by Claude weeks or months later.

This is the step most people skip. And it's the step that creates the compounding effect.

Why the Three Steps Work Together

Each step compounds on the one before it.

CLAUDE.md gives Claude your identity and context. Obsidian + MCP gives Claude your accumulated knowledge. brain-ingest ensures that knowledge keeps growing without manual effort.

Skip one and the others degrade. CLAUDE.md without Obsidian means Claude knows who you are but can't access what you know. Obsidian without brain-ingest means your knowledge base stops growing unless you manually write notes.

Build all three? Claude stops being a tool. It starts feeling like a collaborator who's been working with you for months.

Getting Started This Weekend

You can set up the entire system in about two hours:

  1. Create your CLAUDE.md file. Spend 30 minutes writing a thorough brief about yourself, your projects, and your preferences.
  2. Install Obsidian (free) and set up the smart-connections and qmd MCP servers. Takes about 20 minutes with the documentation.
  3. Install brain-ingest and feed it your first video or voice memo. Watch it create structured notes automatically.

The productivity difference shows up in your very next Claude session. Instead of spending the first 10 minutes re-explaining your project, Claude already knows everything. You jump straight into the work.

Your AI just got a permanent memory. Use it.

Go deeper

Obsidian + Claude Code: Give Your AI a Persistent Memory

Claude forgets everything when a session ends. Wire up an Obsidian vault as a persistent external brain using MCP, and your AI starts walking into each conversation already knowing your projects, preferences, and open decisions.

Read the white paper →
Ready to stop reading and start shipping?

Get a free AI-powered SEO audit of your site

We'll crawl your site, benchmark your local pack, and hand you a prioritized fix list in minutes. No call required.

Run my free audit
WANT THE SHORTCUT

Need help applying this to your business?

The post above is the framework. Spend 30 minutes with me and we'll map it to your specific stack, budget, and timeline. No pitch, just a real scoping conversation.

Common questions

Frequently asked

What is CLAUDE.md and how does it help Claude remember context between sessions?

CLAUDE.md is a file you create in your home folder that Claude reads automatically at the start of every session. You write your name, project details, preferences, and any information you don't want to explain again. Claude retains this information as permanent context, eliminating the need to re-explain your background or decisions in each new conversation.

How does MCP connect Claude to Obsidian notes in real time?

MCP (Model Context Protocol) uses two servers to bridge Claude and Obsidian. The smart-connections server handles semantic search so Claude understands concepts, not just keywords. The qmd server handles structured queries for retrieving specific data from specific notes. Once connected, Claude searches your Obsidian vault automatically as it works without you manually pasting content.

Why should you name Obsidian notes as claims instead of categories?

Naming notes as claims (like "memory graphs beat giant memory files.md") instead of categories (like "memory-systems.md") lets Claude determine relevance from the title alone before reading the content. This makes semantic search faster and more accurate because the title itself conveys the core idea.

What does brain-ingest do and how many ideas can it extract from a single video?

brain-ingest transcribes audio and video sources locally on your machine, extracts key ideas, and automatically creates structured notes in your Obsidian vault. From a single 90-minute conference talk, it can extract 12 to 18 key ideas and 3 to 5 frameworks, all searchable by Claude in future sessions.

How long does it take to set up the complete Claude second brain system?

The entire system takes about two hours to set up. Spend 30 minutes creating your CLAUDE.md file with a thorough brief about yourself and your projects. Install Obsidian and configure the smart-connections and qmd MCP servers in about 20 minutes. Install brain-ingest and process your first audio or video source to see automatic note creation in action.