# El Cafesito By Mexicanada — robots.txt
# Restaurant website: https://www.elcafesitomexicanada.ca

# ---------------------------------------------------------------
# Default rules for all crawlers
# ---------------------------------------------------------------
User-agent: *
Allow: /

# Legal pages (excluded from indexing)
Disallow: /privacy-policy.html
Disallow: /terms-of-use.html

# AI-only resource — should not be indexed as a regular search result
Disallow: /llms.txt

# Common non-content URL patterns
Disallow: /config/
Disallow: /search/
Disallow: /account/
Disallow: /api/
Disallow: /static/
Disallow: /*?*author=*
Disallow: /*?*tag=*
Disallow: /*?*month=*
Disallow: /*?*view=*
Disallow: /*?*format=*

# ---------------------------------------------------------------
# AI / LLM crawlers (training and answer engines)
# Allow site-wide crawling but keep legal + llms.txt out of indexes
# ---------------------------------------------------------------
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: cohere-ai
User-agent: FacebookBot
User-agent: Applebot-Extended
Allow: /
Disallow: /privacy-policy.html
Disallow: /terms-of-use.html
Disallow: /llms.txt

# ---------------------------------------------------------------
# Google Ads bots — must be allowed everywhere ads run
# ---------------------------------------------------------------
User-agent: AdsBot-Google
User-agent: AdsBot-Google-Mobile
User-agent: AdsBot-Google-Mobile-Apps
Allow: /

# ---------------------------------------------------------------
# Crawl delay for heavy bots
# ---------------------------------------------------------------
User-agent: Baiduspider
Crawl-delay: 10

# ---------------------------------------------------------------
# Sitemap
# ---------------------------------------------------------------
Sitemap: https://www.elcafesitomexicanada.ca/sitemap.xml