loop0
← From the Loop
2026-11-15

I was the framework

How my prompts turned into loops, why every loop depended on my memory, and why I took the framework out of my head.

Why I built loop0, and why I use it.

At first it was prompt engineering. Finding the right way to ask, refining the request, trying again. People were starting to call that a loop, but really it was a conversation.

Then coding agents changed the shape of the work. An agent could take a clear piece of work, build it, check it and report back, and I could decide what came next. Without really planning it, my prompts turned into loops: one mission, built, verified, written down, then the next one.

The rhythm came naturally. Keeping every loop at the same level did not. Across projects, I kept reaching for the same things: the same way to frame an idea, the same documents, the same checks before calling something done. My repositories became my knowledge base. Start from that repository. Reuse that file. Do it the way we did last time.

I should say I don't come from programming. I come from tech: I understand systems, architecture and how the pieces fit, but I don't write production code all day. With an agent, I don't have to. What I do have to do is hold the whole thing together — and that is where I had to work hard not to let anything slip.

It all depended on me remembering. Did I ask for the tests to run in a real browser? Did we write down that decision? What was left open last time? Each loop was only as good as my memory that day. I was the framework.

So I took the framework out of my head. loop0 is a small server my coding agent connects to. It doesn't write code and it doesn't call any AI. It gives the agent the same structure on every loop: the purpose written down, the standards set, a mission I approve, proper checks, decisions recorded, and what is still open carried into the next loop.

Now I don't stop to wonder what I forgot. loop0 takes care of the structure. I can think about the product.

What changed for me:

  • Clarity. Every project has its why written down, next to the code.
  • Standards. The same ones across projects, sized to each.
  • Picking a project back up. I open it and the devlog tells me where we are.
  • Decisions. Logged, with who decided and why.
  • Vision. The project follows it, loop after loop.
  • Architecture. It stays stable, because it is documented and every change says what it touched.
  • Agents. They don't have to relearn the project every time.

It doesn't make things faster or cheaper. A loop with loop0 does more: it writes things down and it checks more. I still decide what to build and whether a loop is done. The agent builds. loop0 keeps the structure.

I build loop0 with its own method. It is the only way I would trust it.

If you have been the framework too, try it on your next project. Start at loop zero.