Hi community. I just Wondered whether anyone else has come across this issue: I created a complete repo using the ICM structure, it's an agency style business. I've attached a pic of generally what my repo looks like. For the first few weeks it was working really great. I was able to ask it to generate invoices and proposals and contracts and by and large it was deterministic in nature just following the instructions from the context.md files as expected. But then I started to notice a deterioration in the output. And what would happen is if I wanted to add a custom item for let's say a proposal or an invoice, it would then start updating the config files and changing the workflow. Previous templates that were set up as desired were subtly changing over time as I was making requests for particular clients or particular contracts to include particular clauses. And now I'm in a situation where I feel like I'm correcting things more than I should be and my productivity is getting impacted. It seems like the integrity of my framework has deteriorated. I'm obviously a little bit cautious and worried about having to rework my entire repo, but I'm wondering if anybody has experienced this issue themselves and what steps, if any, they have put in to create some guardrails around this. I also feel like over time it's probably helpful to have a workflow that just cleans house full of clutter and unwanted files that tend to materialize over time as I work on the business. Any ideas, thoughts, or input much appreciated. Thank you.

The problem this time: every "can the AI see this data" conversation turns into a policy document or a system prompt telling the AI not to look. Instructions like that are hopes, not guarantees. What I wanted instead was a structural answer — raw data and AI tools never in the same folder, period, provable from the folder structure itself rather than from a document saying so. So: a raw source folder no AI tool is ever pointed at, a plain deterministic script as the only bridge, and an output folder that's the only thing AI gets to read. I added one more piece after reading through another open-source tool's code (AI Airlock) and noticing a gap in my own first version: nothing was checking that the script's redactions actually held in what it wrote out. Now every run verifies its own output against the raw values it touched, and writes a receipt — pass/fail, a checksum, and an explicit line that passing verification isn't the same as a human clearing the data for use. That part stays a real decision, made by a person, recorded separately. Built it out properly last night — worked examples, tests, a scaffolding script, CI. First CI run caught two real Windows-only bugs I wouldn't have found otherwise, which was its own good reminder that "works on my machine" isn't a claim worth much. It's private for now — not ready for me to just drop a public link and walk away from. But I'd like a few people to actually use it on a real case and tell me where it breaks, before I open it wider. If that's you: comment here or DM me your GitHub username and I'll send an invite.

Since yesterday I am struggling with trying to get my Content Studio to work. I previously had a very simple content studio that would generate content based of the brand voice I had uploaded and the clever prompting and set of predefined list of topics. This produced good stuff but it was just plain simple me talking without any external knowledge. Now the content studio researches on trending topics and makes sure everything is fresh and relevant. However it’s so difficult to get the Next server to work properly locally and just frustrating. How do you guys build a content studio that uses Local LLM and some clever internet trending topic searching?

I was going through AI acronyms and was reading prompting techniques. Now I want to learn those technques does anyone now any resource to learn prompting techniques with practical examples. (I want to improve my prompting)

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