Giving Claude Code a memory

So it remembers who you are and how you work.

Disclaimer: This guide has been created with Claude Code + NotebookLM.

New to Claude Code? Check out this guide first.

The problem with starting fresh every time

You know how every time you start a new conversation with ChatGPT or Claude, you have to re-explain everything?

“I’m working on a novel.”
“I prefer short responses.”
“Don’t use corporate jargon.”

Every. Single. Time.

What if Claude could just... remember?

That’s what this guide is about. Claude Code has a built-in memory system. You tell it things once — who you are, what you’re working on, how you like to work — and it remembers. Every session. Without you having to repeat yourself.

The memory lives in a simple text file called CLAUDE.md. Don’t let the name intimidate you — it’s just a text file with a specific name so Claude knows to look for it. Think of it as a sticky note Claude reads before every session: “Here’s who this person is. Here’s what they’re working on. Here’s how they like things done.”

“Do I need to know how to code for this?”

Nope.

Here’s what this doesn’t require:

• Any programming knowledge

• Understanding what “markdown” means (it’s just text with some formatting)

• Getting everything perfect on the first try

Here’s what you’ll actually do:

• Open a text editor you already have (TextEdit, Notepad, whatever)

• Write some notes about yourself and your projects

• Save the file with a specific name

That’s genuinely it. If you can write a grocery list, you can create a CLAUDE.md file.

And here’s the thing: messy is fine. A memory file with three bullet points beats no memory file at all. You can always add more later — in fact, that’s the whole point.

Creating your memory file: Four steps

Step 1: Create the file

The quick way (from inside Claude Code):

If you already have Claude Code open in the terminal, just type (if you need a reminder on how to do it, you can find it here):

create a CLAUDE.md file for this project

Claude will ask you some questions and create a starter file. Done.

The manual way:

On Mac:

1. Open TextEdit

2. Go to Format → Make Plain Text (important!)

3. Save the file as CLAUDE.md in your project folder

On Windows:

1. Open Notepad

2. Save the file as CLAUDE.md (make sure “Save as type” is “All Files”, not “Text Documents”)

3. Save it in your project folder

That’s it. You have a memory file.

Step 2: Add the basics

Open your CLAUDE.md and start simple. Here’s a starter template:

# About Me

[One or two sentences about who you are and what you're working on]

# My Projects

[List your main projects — even just names and one-line descriptions]

# How I Like to Work

[A few bullet points about your preferences]

Example:

# About Me

I'm a writer working on a mystery novel and a weekly newsletter. I have ADHD, which means I work best with external structure and gentle accountability.

# My Projects

- Novel: "The Last Garden" — a mystery set in a botanical garden
- Newsletter: Weekly essays about creativity and getting unstuck

# How I Like to Work

- Keep responses concise — I get overwhelmed by walls of text
- Ask me clarifying questions instead of making assumptions
- Remind me to take breaks if we've been working for a while

Step 3: Save and test

Save your file. Then:

1. Open Terminal/PowerShell in your project folder

2. Type claude and press Enter

3. Ask Claude something like: “hey Claude, what do you know about me?”

Claude should reference what you wrote. If it does — it’s working.

Step 4: Let Claude help you grow it

Here’s where it gets good. You don’t have to manually update your memory file every time you learn something new about how you work. At the end of a session, just ask:

review our conversation and suggest updates to my CLAUDE.md with the things worth remembering for the future — show me what you'd add before making any changes

Claude will look back at what you discussed, show you what it thinks is worth saving, and wait for your approval before making any changes. You stay in control — nothing gets added without your say-so.

Things Claude might suggest adding:

• A new project you mentioned

• A preference you expressed (”actually, I prefer bullet points over paragraphs”)

• Context that came up (”I only write in the mornings because afternoons are for family”)

• A tool or workflow you use

Over time, your memory file becomes a rich picture of who you are and how you work — and you barely had to write any of it yourself.

Your first “oh wow” moment

Try this:

1. Add a project to your CLAUDE.md (even something simple like “Working on: a short story about time travel”)

2. Save the file

3. Start a new Claude Code session

4. Ask: “What should I work on today?”

Claude will reference your project without you having to explain it. That’s the magic — context that persists.

A few more things to try

If you have multiple projects:

hey Claude, based on what you know about me, which project have I been neglecting?

If you want Claude to match your voice:

Add a “Voice” or “Writing Style” section:

# My Writing Voice

- Conversational, not formal
- Short sentences
- Okay to be a little self-deprecating

Then ask Claude to write something and see if it matches.

If you want Claude to remember your tools:

# Tools I Use

- Writing: Ulysses
- Notes: Obsidian
- Tasks: Notion

Now when you say “add this to my tasks,” Claude knows you mean Notion.

Where memory files can live

Here’s something useful: you can have memory files at different levels:

• ~/.claude/CLAUDE.md — personal preferences that apply everywhere

• ./CLAUDE.md (in a project folder) — project-specific context

Claude reads all of them and combines them. So you might have:

• A personal CLAUDE.md with your general preferences and communication style

• A project CLAUDE.md with details about that specific project

Pro tip: Start with just one file in your main project folder. Add a personal one later if you want.

When you’re ready for more

There’s more you can do with memory files — linking to other files, creating rules for specific file types, organizing instructions into folders. But that’s for later.

For now, just start with one file. A few bullet points about yourself. See how it changes things.

The whole point is this:

You shouldn’t have to re-introduce yourself every time. Give Claude a memory — and get straight to the work.

You got this.

Quick reference

Create your memory file the easy way (from Claude Code):

create a CLAUDE.md file for this project

Create manually:

• Mac: TextEdit → Format → Make Plain Text → Save as CLAUDE.md

• Windows: Notepad → Save as CLAUDE.md (All Files type)

Where it lives:

• Project-level: ./CLAUDE.md (in your project folder)

• User-level: ~/.claude/CLAUDE.md (for preferences that apply everywhere)

Test that it’s working:

what do you know about me?

Grow it over time:

review our conversation and suggest updates to my CLAUDE.md with the things worth remembering for the future — show me what you'd add before making any changes

Starter template:

# About Me
[Who you are, what you're working on]

# My Projects
[List with brief descriptions]

# How I Like to Work
[Your preferences as bullet points]

---

Found a discrepancy or stumbled upon an unexpected issue? Please me know in the comments below so I can update this guide for future readers. Thank you!

---

This guide is a part of my “Get started with Claude Code for creatives” master guide — check it out here.