# START YOAST BLOCK
# ---------------------------
# Disallow access to plugin directories 
User-agent: *
Disallow: /wp-content/plugins/


# Disallow access to the WordPress includes directory
Disallow: /wp-includes/

# Disallow Cloudflare internals (email-protection obfuscation endpoint)
Disallow: /cdn-cgi/

# Disallow access to potential private areas
Disallow: /private/
Disallow: /tmp/
Disallow: /logs/
Disallow: /drafts/

# Disallow low-value parameter URLs that waste crawl budget
# (gallery / pagination / attachment params -> infinite near-duplicate URLs).
# NOTE: only the query params are blocked, never the /page/N/ archive path,
# which Googlebot needs to traverse to reach articles.
Disallow: /*?page=
Disallow: /*?pg=
Disallow: /*?image_id=
Disallow: /*?attachment_id=
Disallow: /*?album_id=
Disallow: /*?wptouch_switch=
Disallow: /*?wptouch_preview_theme=
Disallow: /*?p=
Disallow: /*?author=
Disallow: /*?ref=
Disallow: /*?replytocom=

# Disallow RSS/comment feeds (never indexed; pure crawl-budget waste).
Disallow: /feed/
Disallow: /*/feed/

# AI crawlers are intentionally NOT blocked here (per site owner) — they may
# crawl the site like any other user-agent.

# Specific rule for MJ12bot (aggressive SEO crawler, not an AI bot)
User-agent: MJ12bot
Disallow: /


# Multiple Sitemaps

# AdSense crawler must see every monetised page incl. gallery ?image_id= URLs (24.09.2026)
User-agent: Mediapartners-Google
Allow: /

Sitemap: https://deavita.fr/sitemap_index.xml
Sitemap: https://deavita.fr/news-sitemap.xml
# ---------------------------
# END YOAST BLOCK