# robots.txt - Periville Chemists
# https://www.perivillechemists.co.uk/robots.txt
# Last updated: 2026-06-25
#
# GEO/LLM reference file: https://www.perivillechemists.co.uk/llms.txt
# AI Policy:              https://www.perivillechemists.co.uk/ai-policy.txt
# Privacy Policy:         https://www.perivillechemists.co.uk/privacy-policy/

# ================================================================
# SECTION 1 - Standard search engine crawlers
# Allow full indexing of public content; block system/admin paths
# ================================================================

User-agent: Googlebot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /feed/
Disallow: /?s=
Disallow: /search/
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: Bingbot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /feed/
Disallow: /?s=
Disallow: /search/
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: Applebot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

# ================================================================
# SECTION 2 - GEO / AI answer engine crawlers (PERMITTED)
# These agents power AI-generated answers and citations.
# Allowed to read all public content.
# ================================================================

User-agent: PerplexityBot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: YouBot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=

User-agent: PetalBot
Allow: /
Disallow: /wp-login.php

# ================================================================
# SECTION 3 - AI model crawlers (NOW PERMITTED)
# All AI crawlers are allowed to read public content.
# ================================================================

User-agent: GPTBot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: ChatGPT-User
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: Google-Extended
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: anthropic-ai
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: Claude-Web
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: ClaudeBot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: cohere-ai
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: CCBot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: AI2Bot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: Diffbot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/

User-agent: Omgilibot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=

User-agent: Bytespider
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=

User-agent: FacebookBot
Allow: /
Disallow: /wp-login.php
Disallow: /xmlrpc.php
Disallow: /?s=

User-agent: Timpibot
Allow: /
Disallow: /wp-login.php

# Spam/scraper bots - still blocked
User-agent: ImagesiftBot
Disallow: /

User-agent: img2dataset
Disallow: /

User-agent: Scrapy
Disallow: /

User-agent: DataForSeoBot
Disallow: /

User-agent: PiplBot
Disallow: /

# ================================================================
# SECTION 4 - All other crawlers (default rules)
# ================================================================

User-agent: *
Allow: /

# WordPress system paths
Disallow: /wp-login.php
Disallow: /wp-json/
Disallow: /xmlrpc.php
Disallow: /trackback/
Disallow: /cgi-bin/

# Dynamic and low-value pages
Disallow: /?s=
Disallow: /search/
Disallow: /feed/
Disallow: /comments/feed/
Disallow: /author/

# Transactional / private pages
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/


# ================================================================
# SECTION 5 - Sitemaps
# ================================================================

Sitemap: https://www.perivillechemists.co.uk/post-sitemap.xml
Sitemap: https://www.perivillechemists.co.uk/page-sitemap.xml
Sitemap: https://www.perivillechemists.co.uk/articles-sitemap.xml
Sitemap: https://www.perivillechemists.co.uk/category-sitemap.xml
Sitemap: https://www.perivillechemists.co.uk/articals_catagories-sitemap.xml
Sitemap: https://www.perivillechemists.co.uk/article_tag-sitemap.xml