The dream diary, four months in

These co-builds are casual. We always have something: a template or a focus, and we go through it, talk about it, test it out. You can download it, check it out, and there's always someone in the room who actually uses that specific template, so you can ask questions and compare notes.

Even if the template itself isn't something you'd use day in, day out, there's always something to learn from each one. This one is dream analysis, so it has a really interesting dynamic: different lenses it analyzes the dreams through.

The download gives you a few files. The analysis reference is the framework: the lenses it analyzes through. And the beautiful part of everything being a text file is that you can just see it. If you don't want anything Jungian in there, you literally delete it. The dream diary file is the protocol side: how it goes through logging a dream, what it does with it, where it saves it. The log is the home of your dreams, with specific templates for capturing one, going deeper, and how it shows up in weekly and monthly reviews.

Then there's the setup guide. When you give these four files to Claude Code, it reads through that and knows how to use all of them. You don't need to tell it, and you don't need to read the files yourself either. You can just ask: what is this and how does it work?

Maren: I asked my Claude today how many dreams I've actually analyzed. Since February, through late pregnancy and postpartum: 46 nights of dreams logged and analyzed. 90 plus individual dream segments, because many nights had three or four different dreams. 80 plus personal symbols built in the library, each with my specific meaning, not a generic dictionary definition. And 25 plus recurring themes actively tracked across time.

Maren: I record them right after waking up, otherwise I lose what I dreamt. I just talk into my iPhone notes with dictation, no structure at all. Then during my morning check-in I put the transcript in and it analyzes the dream from that. Speaking works better than writing: it's faster, more unfiltered, and you keep more details.

Maren: Claude pointed out a clear arc from February to now. Tony Robbins started as an unreachable mentor I was projecting onto. By June he was family and I was giving him advice. Three dreams, four months, full integration of the mentor quality: from out there to mine.

Maren: I also dreamt that a colleague on my team was leaving. He actually left the team a few months later. The overview said the psyche processed his departure before waking life confirmed it. When the news arrived, it landed as recognition, not surprise.

Maren: I analyzed my dreams before I had my Claude system, and it was basically repeating what I told it. With the frameworks in the template, it gives me those different angles, and I always get reflection questions that I like a lot.

Maren: I dreamt the other day that someone gifted me blue shoes. It asked me: what is your association with blue? I told it, and it added a personal symbol. I'm very scared of crocodiles, and I've dreamt about them a few times, so now it knows the symbol because it knows my association. Whenever something new appears, it asks what my relationship with it is.

This is one of the only tools in the system where you do get some kind of analysis. The full system is deeply trained to never analyze or give opinions, only mirror back. Here it interprets, and you can check it and correct it when it doesn't align. That's how it keeps adjusting and gets to know you better.

Your dreams live in your local log, on your own device. Only the part you share with Claude touches Anthropic's servers, handled under your plan's terms and their privacy policy. Claude doesn't browse your machine: what it never reads stays local. Compare that to keeping everything inside a chat app, where all of it lives on someone else's servers.

You're building a local library of your own self and your own interpretations.

Note: lightly edited from the call to stay accurate over time. "The part you share" includes any local file Claude reads. The reach rule and current data handling: the Privacy & GDPR module and Anthropic's privacy policy.

100 percent. We actually have a nutrition and food diary session coming up in a few weeks. I use it both to plan and, like Maren was explaining, for the analysis part, because it's really hard to tell, living day to day in your life, what is actually happening.

One interesting thing from my own cycle analysis: apparently my mood and my productivity are not aligned at all with how they're spoken of in the books I trained it on. I'm super productive when I'm supposed to be at my lowest. When I looked closer, it's that I keep timing big pushes to the absolute worst time of my cycle. Of course I'm productive then: I created circumstances where I have to be. You wouldn't notice these things together if you weren't tracking them together.

It's a local file, so it only gets analyzed when you ask. With apps, or with generic chat tools, you have to store everything on their servers. One of the reasons I wanted my cycle data out of an app was finding out the app was selling it. Anything I don't want my data sold for, I tend to bring here, because at least it stays in a controlled space. Nobody using an app gets asked: how many days do you want your data saved in the cloud?

Maren: I did a case study on cycle tracking apps during my studies. They were criticized for sharing data until they made changes. What I liked from the beginning of setting up my system: most things live on my computer, and I decide how much I share.

If something is really sensitive, you can run just that part through API billing, which runs on Anthropic's Commercial Terms rather than your personal plan. And Maren's idea from this call is a good one: an analyze-today-only mode, where you analyze the dream on the day and keep no history at all. You have that choice here.

Note: lightly edited from the call to stay accurate over time. Current data handling: Anthropic's privacy policy and the Privacy & GDPR module.

Look at the logic of it: there's the analysis framework, the steps it takes when it gets a dream, and the log where it saves. So many things could run on that same skeleton. If you give Claude these files, you can just say: create this similar type of tool for another topic, and it will.

It's very similar to how I use my own tarot framework for readings: the same structure of framework, protocol, and log.

Maren: I want to build something around my vision. Before my baby was here, I was meditating every morning and doing visualization. It's so easy in day to day life to focus on what's around you and lose the focus on where you want to go. I want something that keeps me anchored in that.

It reminds me of future pacing, the NLP exercise: you go into a time in the future, visualize it super clearly, then come back step by step to the present. We could do that exercise on a call, record it, give it to Claude, and have it log the steps. Then when you hit those steps, it's easier to see the journey there.

If you bought the book, you paid for it, you own it, and you're using it for yourself, it's not an infringement of copyright in any way. Where it turns into a different area is selling the exact same framework under the same name. For yourself, you own the information. It's yours to use.

One thing to know: whatever you give your root system as a file, it sticks to much better than fetching from its own memory, which hallucinates a lot more. If Claude pushes back, just say: this is a book I purchased. Audiobooks are the hard ones. Use the e-book version, or scan the pages.

We all have frameworks just sitting on a bookshelf. Feed one in and have it run you through it in your morning or evening ritual, even for a month. At least then it's put into practice.

It started with the question that matters: what do I do with the file I downloaded? Go to your own system in the terminal, start Claude, drag the downloaded folder in, and say: can you please add this to my system. It reads the setup guide, checks what's already in your system, and installs it. The add-ons are built non-destructive: it never destroys what's in your system. It finds what already exists and checks where things go.

It adds one pointer line to your root file, so when you mention a dream, it knows where the diary lives. The files aren't fetched every single time, only when there's a reason to use them. It wires itself in the background.

One member said it out loud: I opened the terminal and freaked out. You're not giving Claude full access to your computer. It's folder-level access that you approve.

When you have the full starter system downloaded, you give Claude Code the folder, start it, and just ask: okay, let's get started, what's next? It will take you through it.

If you ask Claude about Claude, it will not be correct. It will be super out of date. But this system is trained on itself. The system knows itself; Claude doesn't know itself.

If you need something current about Claude, have it go to Anthropic's website and fetch what's happening. If you need to know how your own system works, ask the system.

I asked it: can you explain to me how the memory system works? You can always see where it's reading from, which I really like, because once you know your system, you can tell when it's looking in the wrong place. It answered with the same wording the course uses, the map and the terrain, because the system and the course are trained on the same material.

And everything it described, you can change. If I said: I don't want to record learnings, please delete it everywhere, it would. The system isn't something you force your life into. It's a structure to start with, that you adjust and edit to however you want to use it.

Maren: It would be really cool to offer this as an add-on for my program, for clients who work with dreams or want to. The same with the nutrition and exercise templates, because those will be important for me as well.

I wouldn't sell the whole system, but creating templates and selling the templates you create? Definitely. You can make them standalone project templates and include them in your offerings.

Yes, you can connect it to Google Drive with a connector. The one thing I'd look at: decide intentionally where it connects and when, especially if you have decades of files in there, so you're not loading all of it in.

Setting up your system is the perfect time to audit what you have and let go of what's not up to date. The more material in your system that isn't accurate, the more random the output becomes. It's always more fluent with data that lives in the system itself, but not everything needs to. Notion and other platforms work the same way.