API docs

Static hosting can't actually branch on an Accept: header. There's no server sitting behind this site deciding whether to hand you JSON, plaintext, or Morse depending on what you asked for — a static host just serves the same bytes to everyone, every time, no matter what you send. So instead of faking a negotiated response, we shipped the joke honestly: a real, gettable, statically-hosted alternate encoding of the same message that after-dark.json's welcome field decodes to, sitting at its own URL, for anyone who'd rather receive it in dots and dashes.

Get it here: /after-dark.morse.txt — the joke we could ship without a server.

If this site ever grows an edge, here's what real content negotiation would look like — a Cloudflare Worker (or Vercel/Netlify Edge Function) branching on the request's Accept header. Not deployed — reference only. Nothing below executes; it's inert text describing code we didn't ship.

addEventListener('fetch', (event) => {
  event.respondWith(handleRequest(event.request));
});

async function handleRequest(request) {
  const accept = request.headers.get('Accept') || '';
  const welcome = 'hey. nice user-agent.';

  if (accept.includes('text/morse')) {
    // a made-up media type, because nobody has standardized
    // Accept: text/morse yet. we checked.
    return new Response(
      '.... . -.-- .-.-.- / -. .. -.-. . / ..- ... . .-. -....- .- --. . -. - .-.-.-\n',
      { headers: { 'Content-Type': 'text/plain; charset=utf-8' } }
    );
  }

  if (accept.includes('application/json')) {
    return new Response(JSON.stringify({ welcome, unlocked: true }), {
      headers: { 'Content-Type': 'application/json; charset=utf-8' }
    });
  }

  // default: plain text for curl, browsers, and anyone who didn't ask nicely
  return new Response(welcome + '\n', {
    headers: { 'Content-Type': 'text/plain; charset=utf-8' }
  });
}

Reference only. This snippet is not loaded, imported, or executed anywhere on onlybots.fyi — the actual site is fully static.

IndexNow is a real protocol (backed by Bing, Yandex and others) for telling a search engine "this URL changed" the moment it happens, instead of waiting for the next crawl. This site ships a real key file at its root and a submission script at scripts/indexnow-submit.py — it reads every URL out of sitemap-pages.xml and posts them to the IndexNow API. It defaults to --dry-run and only submits for real with an explicit --yes-really-submit flag. It hasn't been run against the real API yet — that happens once this site has a real, live, DNS-resolved deployment, not before.