How this site is built

I write here about keeping software boring and legible, so it would be strange not to say how this one works. There is not much to it, which is the point.

Framework

Nuxt 4, prerendered to static HTML

Hosting

Cloudflare Pages

Content

Markdown in the repo

Styling

Tailwind CSS

Type

Geist and Geist Mono

Runtime

Node 20

No CMS, no database

Every essay is a markdown file in the repository. Projects, bookmarks and the rest are plain JavaScript objects in a static/ folder. Nothing to log into, nothing to migrate, and nothing that can go down on its own. Publishing is writing a file and pushing it.

Every page is also readable as plain markdown: add .md to any URL. There is an llms.txt and a small MCP server for anything reading this without a browser.

Why build it this way?

Because I have spent years shipping into systems that were already live and could not be taken down to be fixed, and the habit sticks. A site with no database has no database to restore. A page that is a file cannot half render. If something breaks, it breaks at build time, in front of me, rather than in front of a reader.

The one place I have not followed my own advice is measurement: the analytics are heavier than I would like, and the privacy policy is specific about what loads and when.