AI crawler index

What is Claude-User?

Claude-User is Anthropic's fetcher for pages users ask about. Fetches pages when Claude users ask questions that need web content.

At a glance

Operator	Anthropic
Purpose	User-triggered fetch
User agent	Not documented by the operator
Obeys robots.txt	Yes, according to the operator
Published IP ranges	https://claude.com/crawling/bots.json
How to verify it	Anthropic says a source IP on its published list indicates the crawler is from Anthropic.
Official documentation	https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler

Should you block Claude-User?

It visits a page only when a user asks the assistant something that needs it. Blocking it stops the assistant from reading your page in those moments, which costs you citations; and because a person starts each request, robots.txt may not stop it.

Unlike the user agents of OpenAI and Perplexity, Anthropic says Claude-User honors robots.txt. Blocking it may reduce your visibility when users ask Claude to look things up.

How to block or allow Claude-User in robots.txt

Block it everywhere:

User-agent: Claude-User
Disallow: /

Allow it everywhere:

User-agent: Claude-User
Allow: /

Rules are matched by the token on the User-agent line. Test the result with the robots.txt tester before you publish it.

Test Claude-User on your robots.txt

How to let Claude-User through a firewall or CDN

robots.txt only tells Claude-User what it may fetch. If a firewall, a CDN or a bot-protection rule blocks the request first, it never reads the file, so allowing it in robots.txt is not enough.

Allow the IP ranges Anthropic publishes (https://claude.com/crawling/bots.json) rather than the user agent, which anyone can fake.

How to verify that a visit is really Claude-User

Anyone can write Claude-User in a user agent, so the user agent alone proves nothing. Take the IP address of the request from your server logs and check it.

Anthropic publishes its IP ranges as JSON. This command downloads the list and prints True if the IP is in it. Replace 203.0.113.7 with the IP from your logs; Python 3 comes with macOS and most Linux systems.

curl -s https://claude.com/crawling/bots.json \
  | python3 -c 'import sys, json, ipaddress as ip
a = ip.ip_address(sys.argv[1])
ranges = json.load(sys.stdin)["prefixes"]
print(any(a in ip.ip_network(r.get("ipv4Prefix") or r.get("ipv6Prefix")) for r in ranges))' 203.0.113.7

Other Anthropic crawlers

ClaudeBot: AI training
Claude-SearchBot: AI search

Check your own site

The AI crawler checker tells you which AI crawlers your robots.txt lets in, and the log file analyzer shows how often each one actually visits. For the bigger picture, see what AI crawlers are and the full AI crawler index.

What our data says about Claude-User

Figures we measured on real AI answers and websites. Each one links to its study, with the method, the sample and the date.

4.1%

4.1% of the websites AI cites most in Spain block Claude-User from crawling their site, and 6.5% name it in a robots.txt group of its own.

Sample: 678 websites AI cites for three or more of 43 brands in Spain. Source: Mencoro (November 2026), in Which AI crawlers do the websites AI cites let in?