# Robots.txt for ExpurHause - Pet Care Services Website

User-agent: *

# Allow access to main content
Allow: /
Allow: /index.html
Allow: /blog.html
Allow: /article1.html
Allow: /article2.html
Allow: /article3.html
Allow: /article4.html
Allow: /thanks.html

# Allow legal pages
Allow: /privacy.html
Allow: /terms.html
Allow: /cookie-policy.html

# Allow CSS and JavaScript files
Allow: /css/
Allow: /js/
Allow: /images/

# Block specific file types that shouldn't be indexed
Disallow: *.log
Disallow: *.tmp
Disallow: *.backup

# Block any test or development files
Disallow: /test/
Disallow: /dev/
Disallow: /staging/

# Allow search engines to access sitemap
Sitemap: https://expurhause.media/sitemap.xml

# Crawl delay (be respectful to server resources)
Crawl-delay: 1

# Specific rules for major search engines
User-agent: Googlebot
Allow: /

User-agent: Bingbot
Allow: /

User-agent: Slurp
Allow: /

# Block aggressive crawlers or scrapers
User-agent: AhrefsBot
Crawl-delay: 10

User-agent: MJ12bot
Crawl-delay: 10

# Block AI training bots if desired (optional)
# User-agent: GPTBot
# Disallow: /

# User-agent: ChatGPT-User
# Disallow: /

# Cache-friendly directives for static assets
# Note: These are informational as robots.txt doesn't control caching
# Actual caching should be configured at server level