# RetreatExplorer.com — Crawler Rules
# Updated: 2026

# ── Allow all crawlers (default) ─────────────────────────────
User-agent: *
Allow: /
Allow: /guide-yoga-europe.html
Allow: /guide-yoga-usa.html
Allow: /sitemap.xml

# Block internal/admin paths
Disallow: /api/
Disallow: /#dashboard
Disallow: /#admin

# ── Major search engines ──────────────────────────────────────
User-agent: Googlebot
Allow: /
Crawl-delay: 1

User-agent: Bingbot
Allow: /
Crawl-delay: 2

User-agent: Slurp
Allow: /
Crawl-delay: 2

User-agent: DuckDuckBot
Allow: /
Crawl-delay: 1

User-agent: Baiduspider
Allow: /
Crawl-delay: 3

User-agent: YandexBot
Allow: /
Crawl-delay: 2

# ── AI crawlers (explicitly allow for AI assistant indexing) ──
User-agent: GPTBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: CCBot
Allow: /

User-agent: anthropic-ai
Allow: /

User-agent: Claude-Web
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: YouBot
Allow: /

User-agent: Applebot
Allow: /

# ── Image crawlers ────────────────────────────────────────────
User-agent: Googlebot-Image
Allow: /

User-agent: Bingbot-Image
Allow: /

# ── Sitemaps ──────────────────────────────────────────────────
Sitemap: https://retreatexplorer.com/sitemap.xml
Sitemap: https://retreatexplorer.com/sitemap-retreats.xml

# ── Host ──────────────────────────────────────────────────────
Host: https://retreatexplorer.com