# ============================================================
#  robots.txt 
#  Version: 1.01
#  Last Updated: 2026.06
# ============================================================


# -------------------------
# --- Search engines (SEO + AI search) ---

User-agent: Googlebot
User-agent: Googlebot-Image
User-agent: Googlebot-Video
User-agent: Google-InspectionTool
User-agent: Bingbot
User-agent: ChatGPT-User
User-agent: OAI-SearchBot
User-agent: PerplexityBot
User-agent: FacebookExternalHit
User-agent: Twitterbot
User-agent: LinkedInBot
User-agent: WhatsApp
User-agent: DuckDuckBot
User-agent: Slurp
User-agent: YouBot
User-agent: Slackbot
User-agent: Discordbot
User-agent: Telegrambot
User-agent: UptimeRobot
User-agent: Pingdom
User-agent: Applebot
Allow: /
Crawl-delay: 10

# -------------------------
# --- AI training bots  ---

# Claude Search - Επιτρέπεται η αναζήτηση αλλά με "φρένο" 31 δευτερολέπτων
User-agent: Claude-SearchBot
# Claude User - Επιτρέπει στους χρήστες να στέλνουν link σου στο Claude
User-agent: Claude-User
Allow: /
Crawl-delay: 10


# --- AI training bots (blocked) ---
# Claude Training - Μπλοκάρισμα του bot που εκπαιδεύει το AI (βαρύ crawling)
User-agent: ClaudeBot
# Google Gemini Training (≠ Googlebot)
User-agent: Google-Extended
# Yandex — μπλοκαρισμένο (δεν έχουμε ρωσικό κοινό)
User-agent: YandexBot
# Baidu — μπλοκαρισμένο (δεν έχουμε κινέζικο κοινό)
User-agent: Baiduspider
# Amazon Alexa Training
User-agent: Amazonbot
# OpenAI — Training only (το ChatGPT-User επιτρέπεται παραπάνω)
User-agent: GPTBot
# Common Crawl — χρησιμοποιείται από πολλά AI models
User-agent: CCBot
# Cohere AI
User-agent: cohere-ai
# Anthropic — Training (το Claude-Web για browsing είναι διαφορετικό)
User-agent: anthropic-ai
# ByteDance/TikTok — Aggressive + AI training
User-agent: Bytespider
# Diffbot — AI knowledge graphs
User-agent: Diffbot
# Meta AI Training (≠ facebookexternalhit)
User-agent: FacebookBot
# Apple AI Training (≠ Applebot)
User-agent: Applebot-Extended
Disallow: /


# -------------------------
# OTHER BOTS/TOOLS


# Omgili/Webhose — Mass scraping
User-agent: Omgilibot
User-agent: Omgili
# Image dataset collection
User-agent: img2dataset
User-agent: ImagesiftBot
# Timpi
User-agent: Timpibot
# DataForSEO
User-agent: DataForSeoBot
User-agent: AhrefsBot
User-agent: SemrushBot
User-agent: MJ12bot
User-agent: DotBot
User-agent: SerpstatBot
User-agent: BLEXBot
User-agent: AspiegelBot
User-agent: barkrowler
User-agent: MegaIndex
User-agent: SEOkicks
User-agent: seoscanners
User-agent: PetalBot
User-agent: SEBot-WA
User-agent: SERankingBacklinksBot
Disallow: /


User-agent: *
Allow: /
Crawl-delay: 10

# --- WordPress  ---
Disallow: /wp-admin/
Disallow: /wp-includes/
Disallow: /wp-content/plugins/
Disallow: /wp-content/themes/
Disallow: /wp-content/cache/
Disallow: /wp-content/upgrade/
Disallow: /xmlrpc.php
Disallow: /wp-cron.php
Disallow: /wp-login.php
Disallow: /wp-config.php
Disallow: /wp-trackback.php
Disallow: /readme.html
Disallow: /license.txt
Disallow: /.htaccess
Disallow: /.env
Disallow: /.git/
Disallow: /wp-json/
Disallow: /?s=
Disallow: /?p=
Disallow: /search/
Disallow: /author/
Disallow: /tag/

# --- other files ---
Disallow: /cart/
Disallow: /checkout/
Disallow: /my-account/
Disallow: /login/
Disallow: /wishlist/
Disallow: /search/
Disallow: /admin/
Disallow: /catalog/
Disallow: /system/
Disallow: /install/
Disallow: /download/
Disallow: /*?sort=
Disallow: /*?order=
Disallow: /*?limit=
Disallow: /*?filter_
Disallow: /index.php?route=account/
Disallow: /index.php?route=checkout/
Disallow: /index.php?route=common/cart/
Disallow: /index.php?route=product/compare/
Disallow: /catalog/view/theme/
Disallow: /import/

# --- System  ---
Disallow: /phpmyadmin/
Disallow: /pma/
Disallow: /cgi-bin/
Disallow: /webalizer/
Disallow: /error_log
Disallow: /phpinfo.php

# Block infinite event calendar URLs
Disallow: /*tribe-bar-date=
Disallow: /*tribe-bar-search=
Disallow: /event/
Disallow: /events/

# --- Crawl delay για άγνωστα bots ---
Crawl-delay: 10