# Royal Punjabi Restaurant & Caterers - robots.txt
# https://www.royalpunjabirestaurant.ca

# ============================================================
# Default rules for all crawlers (incl. search engines)
# Search engines fall under this group and are blocked from
# /llms.txt so it stays out of their index. AI crawlers are
# granted explicit access to /llms.txt in their group below.
# ============================================================
User-agent: *
Allow: /
Disallow: /components/
Disallow: /llms.txt
Disallow: /*?*author=*
Disallow: /*?*tag=*
Disallow: /*?*month=*
Disallow: /*?*view=*
Disallow: /*?*format=*

# ============================================================
# AI / LLM crawlers - full access, including llms.txt
# ============================================================
User-agent: GPTBot
User-agent: ChatGPT-User
User-agent: OAI-SearchBot
User-agent: CCBot
User-agent: anthropic-ai
User-agent: Claude-Web
User-agent: ClaudeBot
User-agent: Google-Extended
User-agent: PerplexityBot
User-agent: PerplexityBot-User
User-agent: cohere-ai
User-agent: FacebookBot
User-agent: Applebot-Extended
Allow: /
Allow: /llms.txt

# ============================================================
# Google Ads bots - full access
# ============================================================
User-agent: AdsBot-Google
User-agent: AdsBot-Google-Mobile
User-agent: AdsBot-Google-Mobile-Apps
Allow: /

# ============================================================
# Heavy crawler - throttle
# ============================================================
User-agent: Baiduspider
Crawl-delay: 10
Disallow: /llms.txt

# ============================================================
# Sitemap
# ============================================================
Sitemap: https://www.royalpunjabirestaurant.ca/sitemap.xml