Technical Publishing Workflow

GLCO now publishes articles from typed Markdown. Each file carries its title, dates, evidence level, image record, and primary sources. A deterministic check blocks common writing and linking errors before the site builds.

The workflow starts with evidence

The author defines the reader's question and chooses an evidence label: tested, sourced, or mixed. Primary sources are gathered before the main draft. Project claims stay separate from GLCO results.

One command creates the article shell

The repository command creates a Markdown file with the required metadata and section order. The draft starts as private until its review fields and sources are complete.

Automated checks catch repeat errors

The content check reviews required metadata, duplicate titles, heading order, visible sources, internal links, image text alternatives, prohibited wording, and reading ease. It warns when sentence structure is too hard to read.

The check does not prove that a claim is true. A person still checks the source and the meaning.

The page is built for readers and crawlers

Nuxt Content turns Markdown into visible HTML. Each article gets a canonical URL, table of contents, organization author, publication dates, Article data, related links, and social image.

What was tested

Four launch articles pass the same metadata and writing check.
Three Systems Lab entries use the same content model.
Internal links are checked against known public routes.
The production build runs the content check before compiling the site.

The reusable instructions live in the repo-local glco-technical-article skill. The next article can follow the same path without copying an old page.

AI assisted with the workflow and draft. GLCO wrote the rules, checked the output, and owns the published decisions.

Sources

Next action

Create the next draft with the repo article command, add primary sources, and run the content check before review.