User-agent: *
#Disallow: /login/

# Block bot/scraper probes that 404 in GSC (no Next.js, no WordPress on this site)
Disallow: /_next/
Disallow: /wp-content/
Disallow: /wp-admin/
Disallow: /assets/img/

# Block on-site search and any other query-driven URL we have no reason to
# index. Our search results are noindex,follow already, but those URLs still
# burn crawl budget and clutter "Crawled - currently not indexed" reports —
# including the literal JSON-LD SearchAction template `?q={search_term_string}`
# that bots crawl when they fail to substitute the placeholder.
Disallow: /search
Disallow: /*?q=
Disallow: /*?s=
Disallow: /*?query=
Disallow: /*?postcode=
Disallow: /*?search=
Disallow: /compare?q=
Disallow: /compare?query=

# Cloudflare's own email-obfuscation endpoint. It is injected into our markup by
# CF, is not a page, and 404s for Googlebot — it surfaced in "Not found (404)".
Disallow: /cdn-cgi/

# Speakable LD-JSON xpath values (/html/body//h1) get crawled as URLs by Google
# and 404 — they are schema fields, not links. Keep them out of the crawl.
Disallow: /html/

# Block the transactional apply/redirect endpoint. It is a GET that fires an
# external affiliate redirect + records a click-lead, so crawlers walking these
# links generate junk leads and invalid affiliate clicks. Never meant to be
# indexed. (Per-bot groups below also need this — Allow:/ overrides * entirely.)
Disallow: /energyquote/apply

Allow: /*.css$
Allow: /*.js$

# Point crawlers to all three sitemaps
Sitemap: https://www.energyplus.co.uk/sitemap.xml
Sitemap: https://www.energyplus.co.uk/sitemaplocal.xml
Sitemap: https://www.energyplus.co.uk/sitemaptariffs.xml

# Optional: hint AI and advanced crawlers
# (OpenAI's GPTBot, Google-Extended, Perplexity etc.)
# Each still gets the apply-endpoint disallow — a more-specific UA group does
# NOT inherit the * group's rules, so the Disallow must be repeated here.
User-agent: GPTBot
Disallow: /energyquote/apply
Allow: /

User-agent: Google-Extended
Disallow: /energyquote/apply
Allow: /

User-agent: CCBot
Disallow: /energyquote/apply
Allow: /

User-agent: ClaudeBot
Disallow: /energyquote/apply
Allow: /

# End of file