I just can't keep this one to myself. When your conversation is getting close to being compacted. Don't let it! I created a hook that warns me I'm at 75%. Then I apply a skill that I call "loop context". /loopcontext (I apologize this is an image and not text. I'm currently on mobile and not going to type all of this out) I've ran several tests against this, where I'm testing the collapsed versions the same contexts versus just looping the context. The improvements are nuts from what I can tell. When I need more info that it needs to reach for. I'll instruct it to grab what it needs then activate the skill again. This way I can continue the conversation and the conversation is seemingly get much better. Let me know what you think!

Hello everyone, i m new here and since the end of my degree (comptability) I m really interested in AI et Angentic system, i m currently following the course too, but I had a question on the path I m going into, My objective is to become AI/Agentic engineer to be able to create automatic system for myself or for company and try to live with it, honestly i asked Claude a week ago how to do it in autodidact and he gave me a strong roadmap, i just want to know if its really pertinent and if any of u have some advice for my learning : 1) Learning Python (basic like being able to read it / debug / understanding environement / HTTP and JSON request / little project) 2) API LLM (prompt system, how API work, how to use it with JSON, make him use tool in a loop, manage API error and cost per call) 3) Agent & MCP (WorkFlow vs Agent, agentic loop, writing an MCP, real project) 4) Prod & Evals (Git, deploy something that work without me, Logs & monitoring, Evals to test and evaluate quality) And voilà that is my whole plan for the next 6 month, so if you have advice or see things that aren't coherent or other things to really learn feel free to tell me i take everything to learn the maximum and be able to really do something in some month 😄