# beardrive.ai
#
# The default here is YES. Answer engines are a distribution channel, not a
# threat to it: a page that GPTBot or ClaudeBot can't fetch can't be cited, and
# a citation is the one link a reader clicks after asking a question. The
# training-vs-retrieval distinction is deliberate — both are allowed, because
# being the source the model already knows is the same win as being the page it
# fetches today.
#
# The app lives on this same origin (beardrive.ai is config.DefaultServer), so
# the disallows below are the hub's own routes: private files, the sync API,
# auth pages. They are gated server-side too — this just keeps crawlers from
# wasting requests on 401s.
#
# There are deliberately NO per-bot groups. Under the robots protocol a crawler
# obeys only its most specific matching group, so a `User-agent: GPTBot` block
# saying `Allow: /` REPLACES the rules below rather than adding to them — which
# silently un-disallowed /s/ and handed every share token to the training
# crawlers. The wildcard group already allows everything a named group would.

User-agent: *
Allow: /
Disallow: /api/
Disallow: /auth/
Disallow: /p-
Disallow: /s/
Disallow: /join/
Disallow: /billing

Sitemap: https://beardrive.ai/sitemap-index.xml