# robots.txt for smklog.com
# Content signals: search allowed, AI answers/citation allowed, AI training
# allowed (owner decision 2026-08-07: being in the training corpora is
# long-term visibility for a small brand; edge blocks are off to match).

User-agent: *
Content-Signal: search=yes, ai-input=yes, ai-train=yes, use=reference
# No "Allow: /" here on purpose. Everything not disallowed is allowed by
# default, so the line added nothing — but it sat above the Disallow rules,
# and a parser that takes the first match rather than the longest one (RFC
# 9309 requires longest; older parsers do not) let it win over every internal
# screen below. Removing it makes the rules hold under either reading.
# Internal screens. Pages 308-redirects the .html form to the clean URL, and
# robots matches prefixes literally, so a rule written against
# /assets/orders.html sat on the far side of a redirect the crawler never
# reaches, and /assets/orders was served to everyone. Extension-less prefixes
# cover both forms, the way /offline already does.
Disallow: /assets/orders
Disallow: /assets/account
Disallow: /assets/console
Disallow: /assets/owner-analytics
Disallow: /assets/label
Disallow: /assets/status
Disallow: /chrome-extension/
# Cloudflare rewrites mailto: links to this path; the bare URL has no decode
# fragment, so a crawler that follows it always gets a 404. Nothing to index.
Disallow: /cdn-cgi/l/
# The PWA offline shell. AI answer bots found it through sw.js and made it the
# single most-retrieved page for answers (40 of 378 in one day) — an empty
# fallback with nothing to cite. Content lives everywhere else.
Disallow: /offline
Disallow: /offline.html

Sitemap: https://smklog.com/sitemap.xml