User-agent: *
Allow: /

Sitemap: https://slidefast.io/sitemap.xml

# Authenticated app routes. These serve the SPA fallback, whose static HTML is
# the prerendered homepage, so without a Disallow a non-rendering crawler reads
# them as duplicates of the homepage. Every one of them also carries
# <meta name="robots" content="noindex"> once the app boots
# (see src/composables/useRouteSeo.ts); this is the belt to that pair of braces.
#
# NOTE: /assets is deliberately NOT listed. It is both a dashboard tab and the
# directory Vite emits every JS and CSS bundle into, and disallowing it stops
# crawlers rendering any page that is not prerendered. The tab is excluded by
# the noindex meta alone.
Disallow: /dashboard
Disallow: /plan
Disallow: /presentations
Disallow: /editor
Disallow: /settings
Disallow: /themes
Disallow: /brand
Disallow: /leads
Disallow: /emojis
Disallow: /batch
Disallow: /onboarding
Disallow: /generating
Disallow: /oauth
Disallow: /export
Disallow: /etudiants/nouveau

# Share links (/s/*, /c/*) are intentionally left crawlable so the noindex meta
# on them can actually be read: a disallowed URL cannot be de-indexed, only
# ignored, and some of these are already in the index.

# ---------------------------------------------------------------------------
# AI crawlers: no restrictions beyond the app routes above.
#
# Cloudflare's "Managed robots.txt" used to prepend a block disallowing every
# AI agent, which made SlideFast uncitable in ChatGPT and Claude however good
# the copy or llms.txt was. That toggle is off and no agent is singled out
# here, so they all fall through to the `*` group: every public page, none of
# the app routes.
#
# Training crawlers (GPTBot, ClaudeBot, CCBot, meta-externalagent) are allowed
# on purpose. The pages they reach are marketing copy, pSEO landings and API
# docs, i.e. text whose whole job is to travel; there is no corpus here worth
# withholding. A model that has read the site can name SlideFast without
# needing to retrieve it, which outlasts any single search result.
#
# Deliberately no named `User-agent` group for the citation agents either: a
# named group replaces the `*` group wholesale rather than extending it, so an
# explicit `Allow: /` would hand them the app routes disallowed above.
# ---------------------------------------------------------------------------