<?xml version="1.0" encoding="UTF-8"?>
<!--
  Three URLs, because the site has three indexable pages: the house page, Legal
  and Privacy. /gone answers 410 Gone and /404 answers 404, both under
  X-Robots-Tag: noindex, follow; a sitemap that submits a URL it is trying to
  have removed is arguing with itself, so neither is listed. The ten retired
  Gulf-era paths redirect to /gone and are likewise absent — a sitemap lists
  destinations, not redirects.

  NO lastmod, AND THAT IS THE POINT. The rule written here used to be "the date
  the file behind each URL last changed, taken from its last commit, not the
  date this sitemap was edited". It was unmaintainable by hand and it did not
  survive: all three entries read 2026-09-07, which was the last commit date of
  THIS FILE, while index.html, legal.html and privacy.html had all moved on to
  2026-09-09 — precisely the mistake the rule forbade, made in the rule's own
  file two days after it was written. There is no build step here to generate
  the value, so the choice is a date that lies or no date at all. A sitemap with
  no lastmod is valid and simply says "unknown"; a sitemap whose lastmod moves
  for unrelated reasons teaches the crawler to stop believing it, which costs
  more than it ever paid. If a generator is added later, the value it must emit
  for each URL is the last commit date of the file behind that URL, and never
  the last commit date of this one.

  NO image entries, for the same reason. Four <image:loc> URLs were listed here
  — coast.jpg, villa.jpg, still-life.jpg and sunday-table.jpg — described as
  "the .jpg each <picture> falls back to". The cinematic redesign moved every
  photograph on the house page to .webp served directly from <img src>, and no
  page has referenced any of those four URLs since; two of them survive only as
  rewrite destinations in vercel.json for legacy addresses. So this file spent
  that period submitting four images that appear on no page. Every photograph on
  the house page is a plain <img> in the served HTML with its alt text beside
  it, which is how a crawler finds an image regardless; a hand-kept parallel
  list adds nothing and, as demonstrated, goes silently wrong at each redesign.
-->
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>https://novrafoods.com/</loc>
    <changefreq>monthly</changefreq>
    <priority>1.0</priority>
  </url>
  <url>
    <loc>https://novrafoods.com/legal</loc>
    <changefreq>monthly</changefreq>
    <priority>0.5</priority>
  </url>
  <url>
    <loc>https://novrafoods.com/privacy</loc>
    <changefreq>monthly</changefreq>
    <priority>0.4</priority>
  </url>
</urlset>
