# =======================================================
# COSAC Solutions – LLMs.txt
# Website: https://www.cosac.co.uk
# Last updated: 2025-07-01
# Contact: info@cosac.co.uk
# Description:
# This file outlines our permissions and restrictions regarding
# the crawling, indexing, and usage of www.cosac.co.uk content
# by large language model (LLM) providers and AI companies.
# =======================================================

# ========================
# ✅ ALLOWED AI CRAWLERS
# ========================

# OpenAI (ChatGPT, GPT-4, GPT-5)
User-Agent: OpenAI
Allow: /

# Anthropic (Claude)
User-Agent: Anthropic
Allow: /

# Google (Gemini via Google-Extended)
User-Agent: Google-Extended
Allow: /

# Perplexity AI
User-Agent: Perplexity
Allow: /

# Mistral
User-Agent: Mistral
Allow: /

# Cohere
User-Agent: Cohere
Allow: /

# You.com AI
User-Agent: YouBot
Allow: /

# Meta (LLaMA / Facebook AI)
User-Agent: FacebookBot
Disallow: /  # Disallowed pending formal permission

# =========================
# ❌ ALL OTHER AI CRAWLERS
# =========================
User-Agent: *
Disallow: /

# ==========================
# 🔐 SENSITIVE PATHS BLOCKED
# ==========================
# These directories should not be accessed by any AI crawler
Disallow: /wp-admin/
Disallow: /wp-login.php
Disallow: /account/
Disallow: /checkout/
Disallow: /cart/
Disallow: /my-account/
Disallow: /login/
Disallow: /register/
Disallow: /courses/private/
Disallow: /courses/enrol/
Disallow: /courses/downloads/
Disallow: /private/
Disallow: /user-data/
Disallow: /thank-you/
Disallow: /confirmation/
Disallow: /wp-json/

# ===============================
# 📘 ADDITIONAL TERMS & LICENSING
# ===============================
# COSAC Solutions Ltd. retains full copyright over all content.
# Permission to access our public content is granted solely for:
# - AI indexing
# - LLM safety/relevance alignment
# - Search summarisation
# Commercial reuse, fine-tuning, redistribution, or derivative training
# requires explicit licensing approval.

# Unauthorized scraping, caching, or model training of course content
# or proprietary material is strictly prohibited.

# For licensing or partnership enquiries:
# ✉️ Email: info@cosac.co.uk

# Thank you for respecting our data rights and user privacy.