Original post: https://www.skool.com/cliefnotes/folder-architecture-applications-outside-of-claude-code?p=0bf6ea8a Update for anyone following this thread — I figured out the auto-load problem. The short version: I was thinking about this wrong. I kept looking for a Cowork equivalent to a Claude Code hook — some way to force a script to run at session start. That doesn't exist, and I don't think it's coming. So I stopped looking for automation and looked at what Cowork does guarantee instead. Here's the one thing it guarantees: the project-level CLAUDE.md always loads at session start. No exceptions, no trigger phrase, no memory required. That's the only file in the whole architecture Cowork reads without being asked. So the fix isn't a workaround at the session level — it's a rebuild at the file level: CLAUDE.md carries zero content. It's routing only. Its job is to tell Claude what to read next, not to hold any of the actual context. The first instruction inside CLAUDE.md is the load command itself — "every session loads three files automatically: CLAUDE.md, CONTEXT.md, PROTOCOLS.md." Because Claude reads CLAUDE.md the moment the session starts, it hits that instruction immediately and pulls in the other two files on its own. CONTEXT.md is the live state — what I'm working on, what's active, what changed since last time. PROTOCOLS.md is the always-on guardrail layer — the behavioral rules that need to be active before a task starts, not loaded in reaction to one. The load logic lives inside the one file the platform already reads. Not in a habit I have to sustain, not in an opener prompt I have to remember to type. The architecture does the remembering. Here's the structure, so the load order actually makes sense: [name]-os/ # An individual's Second Brain operating system as a folder ├── CLAUDE.md # You are here. Routes — never holds content. 60 lines max

🧠 This round was about what a second brain actually is: a context layer you and your AI both read, not a notes app. The Afternoon Tea is the teaching. The High Tea is the room putting it to work on scale, memory, trust, and security. Here is what I want you to understand about these drops, because it is the whole point of being in here. While the videos are valuable and being able to sit and answer your questions is a big reason for them that's not the only value they hold. 📄 Every drop is a set of working files. Markdown built to be used and reused. Each one ends with the exact data to give your AI for your own situation. This round also ships a starter folder you can open, run the self-audit on, and walk away with the skeleton of your own second brain in a sitting. 🤖 I build them expecting you to feed them to your AI. That is the design. Hand a whole round to Claude in a few minutes, whether or not you made it live. The room's thinking is in the files, so you lose almost nothing by catching it later. 🔄 They adapt. A prompt pack is frozen. These are meant to be reshaped: update the context, swap in your own work, bend the templates to your process. And they grow on my side too, as we learn together in these calls. The call is dialogue. The package is that dialogue, crystallized into something you can run. Next round builds on this one. ☕ Afternoon Tea 6 →Afternoon Tea 6 (Second Brain Chat) 🫖 High Tea 10 → High Tea 10 (Second Brain Deep Dive) 🧭 How you should use these: 🔹 Show up live when you can. Your questions shape the next drop. 🔹 When you can't, rewatch, or drop the files into your AI and run the prompt at the bottom. 🔹 Open the starter folder and build your own version. Rename it to your work. It is yours to keep. 📚 A mastermind ends when the call ends. What you get here keeps working after: a structured version of your own thinking (and some of my own thinking!) that improves every round. In my opinion that is worth more than the hour in the room. (or three as some of you stick around in these calls to chat)

Hi Jake & Community — Working through your folder architecture and markdown file methodology and genuinely seeing the value. The context load sequence, the living CONTEXT.md approach, the layered structure — it maps well to how I think about systems. Here's my constraint: I'm working inside Cowork & Chat without access to Claude Code. No terminal, no hooks, no automatic file loading on session start. The files are there. The structure is there. But Cowork doesn't appear to auto-load them at the start of a session — Claude only reads what it's explicitly pointed to. So the architecture exists but the context doesn't load unless I remember to trigger it manually, which defeats part of the purpose. My question: Has anyone implemented this approach in an environment without Claude Code — no CLI, no hooks, no programmatic session triggers? And if so, what's the workaround for the auto-load problem? Specifically wondering: - Is there a reliable way to trigger context load at session start without Claude Code hooks? - Is a "load my context" opener prompt the current best practice for non-Code environments? - Are there tools or patterns that approximate the Claude Code experience for desktop/non-developer users? I'm working on making the case to my org for this approach — so understanding the real implementation path for non-technical users matters a lot. Also, being in a highly restrictive, governed, and large enterprise company creates so many bottlenecks for this approach, but the value and impact of application outweighs the "risks." TYIA, Colin

Hey @Jake Van Clief and everyone, I posted a question about how to implement ICM in CoWork as my organization only allows access to Cowork and not Claude Code, so the auto-hooks was a tricky thing to figure out but I think I did it. Below is the step by step process I've implemented on my work machine that mimics ICM in Claude Code nearly identically. Please go try it out and let me know if this worked for you or not. Here's the full detailed version: For everyone trying to run ICM in Cowork without CLI — here's the complete working setup from scratch, in order. Step 1 — Create your OS folder on Desktop: Open Finder, navigate to your Desktop, create a new folder. Name it [YourName]-OS. This is your permanent AI workspace. Desktop works because enterprise OneDrive syncs it automatically — you get cloud backup without any path complexity. Step 2 — Add your three core files: Place your IDENTITY.md, CONTEXT.md, and REFERENCES.md inside the folder. These are your three context files — identity, current state, and intellectual library. If you don't have them yet, you'll build them after setup. Step 3 — Point Cowork to your folder: Open Claude desktop → Settings → Cowork → Cowork files → Change → navigate to Desktop → select [YourName]-OS. This tells Cowork where your workspace lives. Step 4 — Add as a trusted folder: Still in Settings → Cowork → Trusted Cowork folders → Manage → Add folder → select Desktop/[YourName]-OS. This means Cowork connects without asking for confirmation every session. Step 5 — Set Global instructions in General settings: Settings → General → Global instructions → Edit. Paste this: "At the start of every session, read IDENTITY.md, CONTEXT.md, and REFERENCES.md from the connected folder in that order. Confirm when loaded." Step 6 — Set Global instructions in Cowork settings:

Folders, not frameworks: how Taurus makes Claude repeatable for a whole team Giving an AI agent the right context at the right moment is still the hardest part of using coding agents like Claude Code in real, daily work. We've all felt it: the agent is brilliant when it knows where it is, and frustrating when it doesn't. So how do you give it that context — reliably, for more than one person? A small team will work but what happens when you try to on-board 100+ people? The popular answers don't scale. Elaborate memory systems help a single power user, but in an enterprise they become a liability: they're hard to curate, easy to pollute, and brittle the moment you add more people and more projects. And anything built around one person's bespoke setup — their servers, their wiring, their mental model — is expensive to onboard a whole team onto. Honestly, the wheel hasn't been invented yet. Nobody has a clean, proven answer for how context should work at enterprise scale. This is where the Interpreted Context Methodology (ICM) changes the conversation. Its core idea is deceptively simple: folder structure as agent architecture. Instead of orchestration code or a sprawling memory store, the context lives in the folders themselves. A workspace is just numbered folders for each stage, with markdown files (CLAUDE.md, conventions, reference material, working artifacts) that load in layers when an agent starts there. The agent reads downward and stops when it has enough — typically 2–8k tokens instead of 30–50k. You "configure the factory, not the product": set the workspace up once, then every run reuses it with new inputs. Outputs are plain text, editable, reviewable at every step. ICM is elegant because it's filesystem-native and human-readable — a non-developer can reshape a workflow by moving files. But it has one practical dependency that's easy to overlook: When you add more and more folders agents begin to skip information. Guidelines are missed, rules are overlooked. What worked for one person doesn't work for another because the model scans economically and thinks it knows enough. The solution is again simple, the agent has to actually start in the right folder. Start in a central place and the layered context never loads; start in the right place and the agent is instantly grounded. In a team, "just cd to the correct directory" is exactly the kind of invisible, error-prone step that breaks repeatability.