# Aladin's Pizzeria & Shawarma — robots.txt
# Last Updated: 2026-04-30

# All crawlers
User-agent: *
Allow: /
Disallow: /admin/
Disallow: /config/
Disallow: /search/
Disallow: /api/
Disallow: /static/
Disallow: /logs/
Disallow: /components/
Disallow: /*?*
Disallow: /*&*
Disallow: /privacy-policy.html
Disallow: /terms-of-use.html
Disallow: /llms.txt

# Major AI training / answer-engine bots — opt them out of legal pages
# and the llms.txt file (separate AI manifest, kept out of search index).
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: Perplexity-User
User-agent: Applebot-Extended
User-agent: FacebookBot
User-agent: cohere-ai
Allow: /
Disallow: /privacy-policy.html
Disallow: /terms-of-use.html
Disallow: /llms.txt

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

# Throttle aggressive crawlers
User-agent: Baiduspider
Crawl-delay: 10

User-agent: SemrushBot
Crawl-delay: 10

User-agent: AhrefsBot
Crawl-delay: 10

# Sitemap
Sitemap: https://aladinspizzeria.ca/sitemap.xml