# ReviewPR.com - Advanced Robots.txt
# Updated: 2025-10-12

# Allow all search engines
User-agent: *
Allow: /
Allow: /img/
Allow: /css/
Allow: /js/
Allow: /fonts/

# Disallow sensitive files and directories
Disallow: /admin/
Disallow: /config.php
Disallow: /config_no_headers.php
Disallow: /db.php
Disallow: /env.php
Disallow: /.env
Disallow: /.env.*
Disallow: /process_order.php
Disallow: /login.php
Disallow: /dashboard.php
Disallow: /logout.php
Disallow: /setup_database.php
Disallow: /import_reviewpr_data.php
Disallow: /checkout.php
Disallow: /private/
Disallow: /tmp/
Disallow: /logs/
Disallow: *.log
Disallow: /backup/
Disallow: /cache/
Disallow: /vendor/
Disallow: /node_modules/

# Specific bot configurations
User-agent: Googlebot
Allow: /
Crawl-delay: 1

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

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

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

User-agent: Baiduspider
Allow: /
Crawl-delay: 5

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

User-agent: facebookexternalhit
Allow: /

User-agent: Twitterbot
Allow: /

User-agent: LinkedInBot
Allow: /

User-agent: WhatsApp
Allow: /

User-agent: Applebot
Allow: /

# Block bad bots
User-agent: AhrefsBot
Disallow: /

User-agent: MJ12bot
Disallow: /

User-agent: DotBot
Disallow: /

User-agent: SemrushBot
Disallow: /

User-agent: MegaIndex
Disallow: /

# Sitemaps
Sitemap: https://reviewpr.com/sitemap.xml
Sitemap: https://reviewpr.com/sitemap-images.xml
Sitemap: https://reviewpr.com/sitemap-news.xml

# Host directive (Google specific)
Host: https://reviewpr.com

# Clean URLs preference
Clean-param: utm_source&utm_medium&utm_campaign&utm_term&utm_content /
