I’m curious how much autonomy everyone gives AI in their workflows. Do you mostly use AI as a tool where you approve each meaningful step, or do you let it make decisions and continue working without checking in with you? I’ve been thinking about where that line should be. For those of you building with AI regularly, how much freedom do you actually give it? What can it do without asking you first, and where do you still require a human decision?

Part of my education journey is to teach also my knowledge folder information. I would like to increase my education rate by creating a pipeline, from a url youtube video -->> aggent analyse and get information-->> create a md file I have everything working my last step that are doing manually is the step 1. So is anyone has something to let an agent read, watch, listen a youtube video ?

Ran a four-stage audit pipeline (my first real ICM build), assuming each stage got a fresh window. Nope. After the full run: 302.8k tokens — 281.3k of it transcript. Everything I engineered (CLAUDE.md, contracts, skills) was 25k. My architecture: 8% of the window. The chatter: 92%. One /clear: down to -282k. Messages: 137 tokens. It picked the run back up from the stage output folders without asking me anything, because my CLAUDE.md tells a fresh session to orient from artifacts, never from the conversation. (Protocol block in the comments if you want it.) So the files hold the state, but to run /clear, I have to do it between the stages? How are y'all automating the clear between stages? By hand at every gate? A wrapper running each stage as its own `claude -p`? Something else? Or does it not matter?

Hi, im making a general marketing agent for small biz package in the form of an installable icm type system for non-technical people to use with their simple harnesses like co-work. To confirm, the idea is that the main functions that many would create a complicated skill for, we instead make as folders correct? Like a video ad Gen agent, including its capabilities is all in the folder as routing an instruction files, not as a skill in the agent skills folder? This has always made more sense to me. I always intuitively understood the benefit of this, did it this way originally, as agent skill limitations and reliability has always bothered me a bit, but I'm struggling to articulate the benefits if I had to explain it to a client, why not make it all a skill. Can somebody help me articulate the benefit difference in a good way, logical easy to understand? It's still to intuitive for me. Also interested to know if there are situations where we would make an agent skill instead of a folder with capabilities and how to identify those situations well? Possibly more the simple things are skills and full on agent functions are folders right, but even all skills/agent xapabilities for that function should be in the folder so maybe we only make skills for small random capabilities not part of primary grouped functions? What is a good line to draw? Or don't we use skills at all? Thanks!

New lesson and Module dropped in The Foundation. Module 0.1: Where All Of This Leads. Main video is also on YouTube. 📍This is one of my favorite videos I have ever made. It pulls the whole methodology together in one sitting and showcases a few tools my team has been building on the back end. The dialogue extraction tool Kay built, the voice-controlled Claude Code setup we ran in a live call, the full content pipeline running from one folder. 📝Watch it no matter where you are in the course. If you are brand new, it shows you where this all leads. If you have been here a while, it shows you what we have been quietly building. 🔗 Module Zero is where I will place new foundational videos as time goes on as well. 0.1: Where All Of This Leads - The Foundation · Clief Notes Comments are open. Curious which part stands out most for you.