Once addons existed, a new problem appeared immediately.
I could copy an existing addon, rename files, replace slugs, change constants, rewrite text domains, update custom post types, rebuild taxonomies, adapt shortcodes, move CSS, move JS, and hope I did not miss a hidden old name somewhere.
Or I could ask AI to do a lot of that work.
At the time, that felt almost free. But I was already wondering how long that would stay true. Even if AI stayed useful, did I really want every new addon to depend on a huge context window and a prompt asking it to perform the same risky transformation again?
Rules instead of repeated improvisation
The scaffolder came from that question.
The goal was not to remove AI from the workflow. The goal was to stop using AI for work that was predictable enough to become tooling.
A new addon should be generated from configuration and templates. The scaffolder should copy only the files that belong to the chosen structure, rename what needs to be renamed, skip what is optional, and generate the repeatable files from a data model: CPT registration, taxonomies, meta boxes, shortcodes, templates, CSS, JS, and loader files.
Over time, the scaffolder becomes a catalog of Lubinik patterns. Every real addon teaches it a little more. When a new kind of CPT, shortcode, metabox, or template becomes common enough, it can become a generation template instead of another manual rebuild.
That is the practical point: save human attention and AI attention for the parts that are actually specific to the project.
