WP Manifestindependent plugin directory
manifest / seo / cv-structured-data

CV Structured Data

WordPress plugin: schema.org JSON-LD for my CV page, plus /cv.md and /llms.txt served with correct media types.

by Mark Biek · github.com/markbiek/cv-structured-data · website

★ 0stars
0forks

Install

No release zip yet. The repository archive installs, but the folder name will carry the branch suffix and updates will not flow:

wp plugin install https://github.com/markbiek/cv-structured-data/archive/refs/heads/main.zip
Readme github.com/markbiek/cv-structured-data ↗

A small WordPress plugin that makes mark.biek.org/cv readable by machines as well as people.

It does four things:

  1. Prints a schema.org Person JSON-LD block in the <head> of the CV page.
  2. Serves /cv.md and /llms.txt from the site root with correct media types.
  3. Points robots.txt at /llms.txt.
  4. Adds print styles so the page saves cleanly as a PDF.

Built for a WordPress.com Atomic site (Business plan or higher). Nothing in it is WordPress.com specific, so it works on any WordPress install.

Why serve the files through PHP

A static cv.md uploaded to the web root comes back as application/octet-stream on WordPress.com Atomic — nginx has no entry for .md in its mime map, and there is no .htaccess to override it. Browsers download the file instead of displaying it, and the type="text/markdown" on the alternate link contradicts what the server sends.

Serving through parse_request lets the plugin set Content-Type itself. It also keeps the files in version control with the code, on one deploy path.

The hook only fires when nginx finds no real file at the path and falls through to index.php, which is the case for both paths here.

Print styles

assets/print.css loads with media="print" and only on the CV page, so it cannot affect the screen render. It hides the site header, footer, navigation and the Jetpack subscription block, drops the block theme's screen width constraint so the content fills the page, sets @page margins, and keeps headings from stranding at the foot of a page.

Link destinations are deliberately not appended with a:after. Most of the link text in this CV is already a URL, so expanding would print every address twice.

The page header and footer on the printout, the URL, date and page numbers, come from the browser's print dialog and not from CSS. Turn them off there before saving the PDF.

Checking the hide list

The hide selectors are written against the live theme's markup, and a class that looks like chrome can turn out to wrap the content. .subscription-block did: the theme puts it on two nested groups and the outer one contains .entry-content, so hiding the bare class printed a blank page.

Run this after any theme update, or after editing the hide list:

python3 tools/check-print-css.py

It fetches the CV page, works out which elements each hide selector matches, and fails if any of them match .entry-content or one of its ancestors. Pass a local HTML file instead of the URL to check a saved copy.

Install

Copy the directory into wp-content/plugins/ and activate it, or point WordPress.com's GitHub deployment integration at this repository.

Configure

Everything lives in cv-structured-data.php:

  • PAGE_SLUG — the page that gets the JSON-LD. Currently cv.
  • SERVED_FILES — root paths mapped to media type and backing file.
  • person_schema() — the schema.org record.
  • assets/print.css — the print stylesheet. Versioned by filemtime(), so an edit busts the cache without a plugin version bump.

The CV content itself is in files/cv.md and files/llms.txt.

Two copies of the CV exist: this markdown file and the WordPress page. They do not sync. Update both.

On the phone number

person_schema() deliberately omits telephone. The number is on the human-readable CV page, but putting it in structured data makes it harvestable at scale, which is a meaningfully different exposure. The line is present and commented out if you want it.

Verify

Validate the JSON-LD at validator.schema.org.

Check the media types after deploying:

curl -sI https://mark.biek.org/cv.md    | grep -i content-type
curl -sI https://mark.biek.org/llms.txt | grep -i content-type

Expect text/markdown; charset=utf-8 and text/plain; charset=utf-8.

If either still returns application/octet-stream, a real file exists at that path on disk and nginx is serving it before PHP sees the request. Remove it.

Note on llms.txt

llms.txt is a convention, not a standard. No major crawler has committed to reading it. It costs one file, so it is worth having, but the JSON-LD block is what actually makes the CV machine-readable — it uses a vocabulary parsers already consume.

License

GPL-2.0-or-later.