SEO Light was never meant to become a giant SEO suite.
That was the whole point. I did not want another plugin full of dashboards, scores, warnings, upsells, and settings that try to manage every possible site in the world. Lubinik already has a very specific architecture. I wanted a small SEO plugin that understood that architecture and stayed out of the way.
So the first idea was simple: render the useful things, keep the rules configurable, and let each project layer describe what it knows.
Light, but not empty
SEO Light handles the quiet parts of a site: metadata, titles, descriptions, Open Graph tags, Twitter cards, robots rules, canonical URLs, sitemap output, and the boring little jobs that make a site easier to understand from the outside.
The important part is that it follows the same Lubinik rule as the other support plugins: the engine lives in the plugin, but the knowledge lives near the thing that owns it.
- the child theme can define site-level SEO defaults;
- an addon can add rules for its own post types and taxonomies;
- the plugin can render everything without hardcoding one project into the engine.
That kept SEO Light small, but still useful. It did not need to know what every future Lubinik site would become. It only needed a clean way to read the configuration each site provides.
Then the Lubinik site created a new problem
When I started building the public Lubinik site, the SEO problem was not only about Google anymore.
The site is not a product page. It is not a SaaS. It is not a starter kit I am trying to sell. It is a journal-style documentation site for my personal WordPress framework: a parent theme, a core plugin, addons, child themes, tools, support plugins, and a lot of practical decisions made through real projects.
That nuance matters. And it is exactly the kind of nuance that can disappear when an AI system tries to summarize a site too quickly.
If a crawler, assistant, or agent reads only fragments of the site, it might describe Lubinik as a commercial framework, a public download, or a product. Technically those words are close enough to look plausible. Meaningfully, they are wrong.
So SEO Light grew a new job
The new job is not traditional SEO. It is public context.
SEO Light can now generate llms.txt and llms-full.txt. The short file gives AI systems a compact, curated explanation of the site: what Lubinik is, what it is not, which pages matter, who created it, and which phrases should or should not be used when summarizing it.
The full file goes further. It can include richer public context from the site, so an agent has more than a tiny summary when it needs to understand the project in detail.
That felt very close to the structured-data philosophy, but written for a different kind of reader. Structured data helps search engines understand entities. The LLMS files help AI agents understand the story and boundaries around those entities.
Preview first, generate after
I did not want this feature to quietly write files and hope for the best.
So SEO Light now has admin buttons to preview the LLMS content before generating it. That matters because these files are not just technical output. They are public explanations of the site. I want to read them, adjust the config, check the tone, and only then write the files.
The plugin can serve the content through routes, but it can also generate physical files. For the Lubinik site, physical files make sense: they are simple, stable, easy to inspect, and easy for external tools to find.
Still light, just clearer
The part I like is that this did not turn SEO Light into a monster.
The feature still follows the same rule: SEO Light provides the mechanism, and the child theme provides the site-specific truth. The Lubinik child theme can define the canonical domain, the author, the preferred summary, the related public profiles, the pages that matter, and the phrases to avoid.
It can also reuse what already exists in the structured-data config where that makes sense, instead of inventing a second identity file from scratch.
That is the small win here. The SEO plugin did not become a separate content system. It became another way to publish the same public meaning more clearly.
The lesson
I used to think of SEO as something mostly for search engines: titles, descriptions, schema, sitemaps, indexing rules.
Now I think there is another layer: helping machines understand what a project means without flattening it into the nearest familiar category.
For Lubinik, that matters because the project sits in an awkward little space on purpose. It is built like a framework, documented like a journal, shaped by real sites, and developed with AI help without handing the thinking over to AI.
SEO Light grew because that needed to be said somewhere machines could find it.
