# robots.txt for mosaya.ai (the same file answers on thesage.ai, where the signed-in app lives)

User-agent: *
Allow: /

# Sitemap: the public pages live at www.mosaya.ai (lib/house-domain.ts)
Sitemap: https://www.mosaya.ai/sitemap.xml

# Internal paths. /_next/ is NOT blocked, on purpose: it holds the CSS, the
# scripts and every next/image photo, and a crawler that cannot fetch them
# cannot render the page or show its images (SEO audit 2026-09-26). No
# Crawl-delay either: Google ignores it and Bing would only slow down.
Disallow: /api/
Disallow: /dashboard/
Disallow: /node_modules/
Disallow: /*.map$

# ===========================================
# AI CRAWLER CONTROLS
# Posture: "cite me, don't silently absorb me."
# ALLOW engines that fetch our pages to cite them live WITH a link back
# (answer/search engines = real discovery + referral traffic).
# BLOCK pure training-only crawlers that give nothing back.
# Revisit per-path when canonical Way of MOS content is published —
# deep philosophy URLs can be training-blocked while marketing stays open.
# ===========================================

# --- ALLOWED: answer / citation engines (they link back to us) ---
# A crawler obeys ONLY the most specific group that names it, so each named
# group repeats the internal disallows above; without them these agents were
# free to fetch /api/ and /dashboard/.

# OpenAI — ChatGPT Search + user-triggered browsing (cites with links)
User-agent: OAI-SearchBot
Allow: /
Disallow: /api/
Disallow: /dashboard/
User-agent: ChatGPT-User
Allow: /
Disallow: /api/
Disallow: /dashboard/

# Anthropic — Claude search + user-triggered browsing (cites with links)
User-agent: Claude-SearchBot
Allow: /
Disallow: /api/
Disallow: /dashboard/
User-agent: Claude-User
Allow: /
Disallow: /api/
Disallow: /dashboard/

# Perplexity — cites sources properly and drives traffic
User-agent: PerplexityBot
Allow: /
Disallow: /api/
Disallow: /dashboard/
User-agent: Perplexity-User
Allow: /
Disallow: /api/
Disallow: /dashboard/

# Google-Extended is dual-purpose: the ONLY switch for both Google AI
# Overviews grounding AND Gemini training. Allowed so we appear in AI
# Overviews, accepting that Gemini may also train on us (intentional trade-off).
User-agent: Google-Extended
Allow: /
Disallow: /api/
Disallow: /dashboard/

# --- BLOCKED: training-only crawlers (no attribution, no traffic) ---

# OpenAI training crawler (ChatGPT Search via OAI-SearchBot stays allowed above)
User-agent: GPTBot
Disallow: /

# Anthropic training crawler (Claude search via Claude-SearchBot stays allowed above)
User-agent: ClaudeBot
Disallow: /

# Common Crawl (heavily used for AI training datasets)
User-agent: CCBot
Disallow: /

# Cohere AI
User-agent: cohere-ai
Disallow: /

# Meta / Facebook AI training
User-agent: FacebookBot
Disallow: /

# ByteDance / TikTok AI
User-agent: Bytespider
Disallow: /

# Amazon AI
User-agent: Amazonbot
Disallow: /

# Apple Intelligence training opt-out (regular Applebot for Siri/search stays allowed)
User-agent: Applebot-Extended
Disallow: /

# Googlebot, Bingbot, Applebot and standard search crawlers remain allowed by default.
# Bing/Copilot answers use Bingbot, so Microsoft Copilot visibility is already on.