Building vanhtran.dev: static, MDX, nothing to log in to
Why the home for this experiment is plain files in a repo, and what that makes easy for crawlers and AI engines.
Before I can ask ChatGPT, Perplexity and Gemini who I am, there has to be one place on the web that answers the question clearly. This site is that place.
What it is
A personal site with three lanes: the Build Log for the 180-day experiment, Writing for essays that should outlast it, and Physical AI for articles on where I'm heading. Every post is an MDX file in a Git repo. There is no database, no CMS and nothing to log in to.
Why static
Every page is rendered to HTML at build time and ships without client-side JavaScript for the interface. Crawlers and AI engines get the full text on the first request, and there is nothing to break at 2am.
The artifact is the credential. The site has to be the most boring, most readable version of that artifact.
The entity work
For stage one of the experiment, the site tells machines who I am in the plainest way I know:
- A
Personrecord in structured data, with my alternate name and links to every public profile. Articlestructured data on every post.- A sitemap, an RSS feed, and an
llms.txtthat explains the site in plain language.