for people just starting, here's what I did to start organizing my file system after a few days of use. I have a little bit of coding background so I asked claude to treat my file system like computer architecture and follow SOLID principles. They are roughly summarized as follows: - Separation of concerns — each file has one job - DRY (Don't Repeat Yourself) — data lives in one place - Single source of truth — no two files own the same information - Load only what you need — frequently accessed files stay shallow and small As example my claude.md file shrunk about 50%, it had grown from 30-40 lines to 100 or so. And the context file in another workspace went from 157 lines to 15. The goal isn't necessarily file reduction, it's efficiency and clarity. I went step by step through each directory with Claude to clean up stuff. At the end I had Claude summarize the process for me to share. attached are pictures of the .md file for easy viewing and I included the actual .md file of the process. But I'll also list it below: The Process Every refactor follows the same pattern: 1. Audit — list every job the file or directory is currently doing 2. Identify what's wrong — duplication, mixed concerns, wrong location 3. Propose before acting — show the before/after and confirm scope 4. Execute in order — move files first, update references second 5. Verify — make sure all links and paths still resolve The key question at each step: does this file have one job, and does it live where you'd naturally look for it? Prompts Used The actual prompts from this session, mapped to each process step. Audit "look at Claude.md architecture. analyse for separation of concerns and good computer architecture" "let's look at sound design context file first" Identify violations "looking at Claude.md Identity, workspaces and routing section. Is there overlap of function there?" "as a computer system architect taking into account separation of concerns, DRY, and clarity, what is the most efficient format?"

Hey All wanted to share something I was able to use today that helped me get something done in 20min rather than the usually 2-3 hours. I like to use Youtube a lot to learn new techniques and ideas to use in my work, but that takes a lot of time. I wanted to learn a specific granular synthesis patch to recreate a sound. Under normal conditions: watch the video, rewatch the confusing parts, look up GRM's modulation system separately, rebuild the patch by trial and error. Realistically 2–3 hours before I'm actually working. With this workflow: clipped the video, had Claude reconstruct the signal flow and give me step-by-step instructions, followed them, asked for timestamps twice when I needed to see something visual. I was building the patch within seconds. Here's the breakdown of the workflow pasted from my session summary from claude: --- ## The Problem YouTube tutorials are great but they also take a lot of time to watch and digest. A 45-minute walkthrough covers maybe 20 minutes of usable information, buried in real-time narration, visual demos, and the host's live troubleshooting. To actually *do* what you watched, you still have to reverse-engineer the steps or follow along. But if I'm hunting YouTube for a tutorial for something I need to do right now that's a time sink. --- ## The Workflow Works for any tutorial, any domain. **Step 1: Clip it.** - Use Obsidian Web Clipper to save the YouTube video as a markdown file. The clipper pulls the transcript and metadata into a structured note directly to an Obsidian vault you can work with directly in a chat. - If you store the Obsidian Vault inside your Claude workspace in a directory you can access it with Claude. **Step 2: Have Claude teach it to you.** - Ask Claude to read the transcript, summarize the technique and walk you through it step by step. - you can immediately start working or learning. **Step 3: Execute with Claude on standby.** - If you need calrifcation ask Claude for the timestamp to a specific step. It can point you back to the exact moment in the source video without you having to scrub through the whole thing. It will even give you a direct link to that point in the video.

The challenge: build a 30-second explainer on any concept. This is where you post it. Where are you at?

Claude design is fully released. It burns up tokens but it works very well, and allows you to automate the folder design process that I've been showing you all. My video will go on to how not to use it as I think there's still some limits to what it can be done, but it's the step in the right direction. Again, I've been telling you about building systems that will be amplified by updates, not replaced. This is a perfect example. I'm going to go ahead and make a nice long form video out about this. What do you want out of the YouTube video? Comment below what use cases you want me to explore in the long-form (and eventually course addition of course !) For those of you that have been studying this the past few months, how does it feel to be ahead for once!?

This is the lesson that pays for itself. Five steps before Claude Code creates a single file. Spend your thinking before you spend your tokens. 💬 Drop Your Answer in the Comments 1. What's a project you're planning to build? Doesn't have to be a website. Internal tool, content system, client work, anything. 2. Write your Step 3 prompt and post it here. Don't run it yet. Let the community give feedback before you burn tokens. 3. If you've already tried this method: how many total prompts from start to finished build? The benchmark is under 10. What would you front-load differently next time? ✅ Action Step Run Steps 1 and 2 right now in Claude chat. Pick any project. Analyze what exists, then ask Claude to write a markdown briefing document for Claude Code to read. Save that file. Drop it in your project folder. You just did the thing that saves the most tokens long term. Quick Gut Check Why do you create a markdown briefing document in Claude chat before opening Claude Code?