# --------------------------------------------------------------------
# SamuiForSale robots.txt  (Joomla 4/5 + JA Justitia + 4SEO)
# Last updated: 2025‑07‑22
# --------------------------------------------------------------------

User-agent: *
# --- Block back‑end and technical folders
Disallow: /administrator/
Disallow: /cli/
Disallow: /installation/
Disallow: /logs/
Disallow: /tmp/
# --- Avoid duplicate component URLs
Disallow: /component/
# --- Allow critical assets for full rendering
Allow: /modules/*.css
Allow: /modules/*.js
Allow: /plugins/*.css
Allow: /plugins/*.js
Allow: /libraries/*.css
Allow: /libraries/*.js
# --- (Optional) block query‑string variants if they appear
# Disallow: /*?search=
# Disallow: /*?filter=

# --------------------------------------------------------------------
# XML sitemap
Sitemap: https://www.samuiforsale.com/sitemap-4seo.xml
# --------------------------------------------------------------------

# =============================
#  AI‑specific directives
# =============================

# --- Block model‑training crawlers ---
User-agent: GPTBot          # OpenAI
Disallow: /

User-agent: ClaudeBot       # Anthropic
Disallow: /

User-agent: CCBot           # Common Crawl (used for datasets)
Disallow: /

# --- Allow AI search‑assistant bots ---
User-agent: PerplexityBot   # Perplexity AI
Allow: /

User-agent: Google-Extended # Google’s AI snippets & SGE
Allow: /

# --------------------------------------------------------------------
# End of file
# --------------------------------------------------------------------