# Default — for all crawlers not explicitly listed below
User-agent: *
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php

# --- AI & LLM Optimization Section ---
# These bots are explicitly allowed to crawl WeDoWebApps content
# for AI training and real-time AI search answers.
# Each block repeats the disallows because robots.txt does NOT
# inherit rules from the wildcard (*) block.

# OpenAI (ChatGPT, GPT-4, GPT-5, ChatGPT Search)
User-agent: GPTBot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: ChatGPT-User
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: OAI-SearchBot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# Anthropic (Claude)
User-agent: ClaudeBot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: Claude-Web
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: Claude-SearchBot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: anthropic-ai
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# Google (Gemini & AI Overviews)
User-agent: Google-Extended
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# Perplexity AI
User-agent: PerplexityBot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: Perplexity-User
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# Apple (Apple Intelligence / Siri)
User-agent: Applebot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: Applebot-Extended
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# Meta (Meta AI / Llama / Facebook link previews)
User-agent: FacebookBot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: facebookexternalhit
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

User-agent: Meta-ExternalAgent
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# Common Crawl (foundation for many open-source LLMs)
User-agent: CCBot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# ByteDance / TikTok / Doubao AI
User-agent: Bytespider
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# Amazon (Alexa / Rufus AI)
User-agent: Amazonbot
Disallow: /wp-admin/
Disallow: /cdn-cgi/
Disallow: /feed/
Disallow: /comments/
Allow: /wp-admin/admin-ajax.php
Allow: /

# --- Sitemaps and Discovery Files ---
Sitemap: https://www.wedowebapps.com/sitemap.xml

# llms.txt files (AI discovery)
# https://www.wedowebapps.com/llms.txt
# https://www.wedowebapps.com/llms-full.txt
# https://www.wedowebapps.com/llms-blogs.txt