Get cited by Claude — Anthropic's assistant — by allowing ClaudeBot in robots.txt, publishing crawlable server-rendered pages with clear attribution markup, and being present in the public web corpora Claude uses for grounded answers.
Anthropic's crawler documentation lists ClaudeBot as its web fetcher, along with related agents like claude-web. Blocking either in robots.txt prevents Claude from including that path when its retrieval features are used.
Claude's web-search feature (available in Claude for Work and via the API's web-search tool) grounds answers on real-time retrieved pages. The same principles that help other answer engines apply: server-rendered HTML, a direct first-sentence answer, Article or FAQPage JSON-LD, and an author entity.
Anthropic does not currently publish a webmaster dashboard for citation impressions. Measurement is manual — query Claude on target questions and inspect the cited sources.
Allow ClaudeBot deliberately; the training-versus-search distinction matters less for Claude than for OpenAI because Anthropic's public documentation groups its crawlers more coarsely. Publishers who want to be citable but not used for training should read the current docs carefully at deployment time.