User-agent: *
# atlas-analytics beacon — no content, and crawlers that render the page were
# spending real crawl budget on it. Keeps the budget on castle pages.
# Listed BEFORE "Allow: /" on purpose: Google resolves by longest match, but
# simpler crawlers take the first matching line, and those are most of the
# beacon traffic. This order blocks it under both readings.
Disallow: /_a/
Disallow: /_a.js
# The map's deep link. Every castle page links its own `/?castle=<slug>` (the
# "view on the map" button, also the TouristAttraction's hasMap), which is
# 9,040 x 8 editions = ~72,320 crawlable URLs that all serve the same
# 8 homepages. Each carries a canonical back to its homepage, so nothing was ever
# indexed twice — but Google had fetched 1,627 of them by 2026-08-21 while
# 23,986 real pages sat in "Discovered, currently not indexed". The budget
# belongs to those. The link keeps working for people; crawlers skip it.
Disallow: /*?castle=
# The same deep link for a country: every country page links its own
# `/?country=<slug>` (8 editions of ~155 of them), each canonical to its
# homepage. Google had 48 of them filed as "alternate page with proper
# canonical" and had served one in results by 2026-10-27 (/?country=romania).
Disallow: /*?country=
# The MCP endpoint (an API for AI agents, see /about/ and llms.txt): a GET
# answers 405, and Google had crawled it ("blocked due to other 4xx", 2026-10-27).
Disallow: /mcp
Allow: /

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