A beginner maker lost time by building inside the main repo
A personal maker with almost no coding background spent three months building a project with AI agents and ran into a painful workflow problem. was created directly inside the real , so unfinished files, experiments, and loose ends kept piling up.
The agents repeatedly warned that the workspace was dirty, which often meant something was incomplete, missing, or not connected yet. A cleaner move to a new had almost solved the problem about a month earlier, but the same messy pattern returned.
The clearer lesson is that it may have been easier to build outside the main first, then move in only the finished and organized parts. The remaining concern is how to keep an AI agent from losing context when the conversation or gets .
Key points
- A non-coder spent three months building a project with AI agents.
- Working directly inside the main caused experiments and unfinished pieces to pile up.
- The agents kept warning that the workspace was dirty.
- A cleaner migration helped once, but the mess returned after the same habits continued.
- Keeping agent context stable is still an open problem for this workflow.