# Robots.txt for Ledsreact - Sports Performance Testing Technology
# Website: https://ledsreact.com
# Generated following SEO best practices

# Default rule for all robots
User-agent: *

# Allow crawling of all public content
Allow: /

# Disallow crawling of technical and admin areas
Disallow: /api/
Disallow: /_next/
Disallow: /.next/
Disallow: /admin/
Disallow: /dashboard/
Disallow: /private/
Disallow: /temp/
Disallow: /tmp/

# Disallow crawling of development and testing files
Disallow: /*.json$
Disallow: /*.log$
Disallow: /*test*
Disallow: /*dev*

# Disallow crawling of sensitive configuration files
Disallow: /config/
Disallow: /.env*
Disallow: /.git/
Disallow: /node_modules/

# Disallow crawling of duplicate content with URL parameters
Disallow: /*?*utm_*
Disallow: /*?*ref=*
Disallow: /*?*fbclid=*
Disallow: /*?*gclid=*

# Allow crawling of important static assets
Allow: /img/
Allow: /video/
Allow: /whitepapers/
Allow: /clients/
Allow: /product/
Allow: /*.css$
Allow: /*.js$
Allow: /*.png$
Allow: /*.jpg$
Allow: /*.jpeg$
Allow: /*.gif$
Allow: /*.webp$
Allow: /*.svg$
Allow: /*.ico$
Allow: /*.pdf$

# Special rules for major search engines
User-agent: Googlebot
Allow: /
Crawl-delay: 1

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

User-agent: facebookexternalhit
Allow: /

User-agent: Twitterbot
Allow: /

User-agent: LinkedInBot
Allow: /

# Block competitive analysis and aggressive scraping bots
# Note: These are legitimate tools but blocked to limit competitive intelligence

User-agent: AhrefsBot
Disallow: /

# Currently allowing this one, for ourselves
#User-agent: SemrushBot
#Disallow: /

# Block known aggressive scrapers
User-agent: MJ12bot
Disallow: /

User-agent: DotBot
Disallow: /

User-agent: ia_archiver
Disallow: /

# Sitemap references
Sitemap: https://ledsreact.com/sitemap.xml

# Host declaration (helps with international targeting)
Host: https://ledsreact.com