# looksmith.ai
#
# Everything is open to everything. This is a marketing site for a pre-launch
# consumer app: being read is the entire point, and there is nothing here worth
# withholding.
#
# This file is live. Until 2026-08-22 it was not: Cloudflare's managed
# robots.txt was on and prepended a block that said Disallow: / to nine AI
# agents, and separately the AI bot policy returned 403 at the edge to
# OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User, PerplexityBot,
# Perplexity-User, GPTBot, ClaudeBot and Amazonbot. Both are now off and all
# nine return 200.
#
# If that ever regresses, robots.txt will not tell you. It is advisory and the
# 403 was enforcement, so the file can look perfect while the edge refuses
# every crawler named in it. The only thing that knows is
# scripts/crawler-probe.sh. Run it after any change in the Cloudflare
# dashboard, and do not trust a panel that says Allow.
#
# One group applies per crawler: the most specific matching User-agent, and it
# inherits nothing from *. If a named group is ever added below, it has to
# repeat Allow: / and the Sitemap: line itself.

User-agent: *
Allow: /

# search:   indexing and linking to this site
# ai-input: grounding a generative answer on this page, live (RAG)
# ai-train: training or fine-tuning a model on this text
#
# All three are yes, deliberately. ai-input=yes is the one that decides whether
# an assistant may quote this site when someone asks about it, and omitting it
# would neither grant nor restrict, which is not a position worth taking.
# ai-train=yes covers the words on this website. It says nothing about user
# photographs, which are governed by the Privacy Policy and by a separate
# written consent, and which no model may train on.
Content-Signal: search=yes,ai-input=yes,ai-train=yes

Sitemap: https://looksmith.ai/sitemap.xml