User-agent: *
Disallow: /admin/
Disallow: /account/
Disallow: /api/
# Account / auth / billing flows (/login/, /signup/, /checkout/, /refund/,
# /lost-password/, /restore-password/, /verify/, /success/, /cancel/,
# /delete-account/) are deliberately NOT disallowed: each one renders
# , and a crawler that is kept
# out by robots.txt never sees that tag, so a linked URL could still be
# indexed from its anchor text alone (portfolio review 2026-09-10).
Disallow: /logout/
# Crawl traps: language-chooser links carry ?return=, login flows ?next=, and
# thin internal search / re-order pages use ?q= / ?sort=. ?lang= stays
# crawlable (it canonicalizes to the English page).
Disallow: /*?return=
Disallow: /*?next=
Disallow: /*?q=
Disallow: /*?sort=
Disallow: /*?ref=
Disallow: /*?from=
# Per-session / per-user results - infinite cardinality, zero index value.
# Paired with in the templates:
# robots.txt alone prevents crawling, NOT indexing.
Disallow: /critique/r/
Disallow: /trace/r/
Disallow: /s/
Disallow: /practice/session/
Disallow: /gallery/
Disallow: /my/
Disallow: /u/
Disallow: /daily/20
# Alias URLs 301 to the canonical short slug and must never be crawled as
# destinations. See docs/SEO-ARCHITECTURE.md S1.1.
Disallow: /how-to-draw-
Disallow: /draw-a-
Disallow: /draw-an-
# Printable step sheets are assets, not pages. They also carry
# X-Robots-Tag: noindex + a canonical header.
Disallow: /dl/
# Commercial SEO-index crawlers. These build backlink databases they sell;
# they never send a visitor, so every request is pure cost - and this is a
# SHARED box, so the cost is partly the co-tenant's.
#
# Measured on 2026-08-11, the first full day this site had an access log at
# all: SERanking made 10,506 requests to 104 distinct URLs in 25 hours. That
# is ~100 re-fetches of every page it knows about, and 70% of ALL traffic to
# draw.how - against 181 requests from Googlebot in the same window. It is not
# crawling, and nothing about this site changes 100 times a day.
#
# The criterion for this list is "does it ever send a human here", not "is it
# rude". Googlebot, bingbot, Applebot and the AI answer crawlers stay welcome
# above; an answer engine that restates a lesson still cannot give anyone a
# canvas, which is the whole AI-Overview defence.
User-agent: SERanking
Disallow: /
User-agent: SemrushBot
Disallow: /
User-agent: AhrefsBot
Disallow: /
User-agent: MJ12bot
Disallow: /
User-agent: DotBot
Disallow: /
User-agent: BLEXBot
Disallow: /
Sitemap: https://draw.how/sitemap.xml