Claude-User: Claude reading your page for a user
Claude-User is the agent Anthropic's Claude sends when a user asks a question that needs a web page. It does not crawl on a schedule. One request in your log means one person, somewhere, asked Claude something your page could answer.
That makes it one of the most useful AI signals you can watch. It tells you which pages people reach for through an assistant, not just which pages a crawler happened to pass.
Claude-User
- Operator
- Anthropic
- Type
- Live fetch for a user
- robots.txt token
Claude-User- Obeys robots.txt
- Yes, by the operator's documentation. Anthropic says the token lets site owners control which sites user requests can reach
- How to verify
- Match the IP against claude.com/crawling/bots.json
- Official docs
- support.claude.com
What a Claude user visit tells you
Anthropic describes the agent simply: when people ask Claude questions, it may access websites under this name. So each hit lines up with a real question.
For example, a reader asks Claude to compare two plugins, and Claude opens your review to check the facts. Or someone pastes your URL into a chat and asks for a summary. Either way, your content lands in an answer.
However, a fetch is not a visit. The person may never click through to your site. The live read tells you that you were consulted, while a referral visit tells you that someone came.
Can you block Claude from fetching for a user?
Yes. Unlike some live fetchers from other companies, this one follows robots.txt. Anthropic says its bots honor robots.txt, and that this token lets site owners control which sites user requests can reach.
User-agent: Claude-User
Disallow: /Anthropic also advises against blocking its IP addresses, because then the agent cannot read your robots.txt at all. So for this agent, robots.txt is both the gentle and the reliable route.
Pagereckon Pro offers one-click blocking for bots that crawl hard and give nothing back, such as Bytespider. Live readers like this one stand for real questions, so the Pro button does not cover them.
The cost of blocking Claude user requests
Anthropic is direct about it. If you disable the agent, Claude cannot retrieve your content when a user asks, and that may reduce your visibility in user-directed web search.
Also note that this rule has no effect on training. For that, you need a separate rule for ClaudeBot. Many sites block training and allow live reads, since only the live reads can lead to a cited link.
Verify a Claude user request
Anthropic's help page names the agent but does not print a full user agent string. To confirm a hit, compare its IP address with the list at claude.com/crawling/bots.json. A match means the request came from Anthropic.
Keep this check in mind before you draw conclusions from a sudden spike. A scraper can borrow the name, and only the IP list tells the two apart.
Claude user reads in Pagereckon
Pagereckon files the agent under live reads, with ChatGPT-User and Perplexity-User. The Assistants card puts Claude's live reads next to the visits that came from claude.ai, so you see both sides at once.
Next, open a page that Claude reads often. The admin-bar panel shows its own AI reads for the last 24 hours. Then check the AI crawler checker and the AI referral traffic report to see if those reads turn into visits.
Claude user agent questions
Is Claude-User a crawler?
No. It fetches a page only when a Claude user asks a question that needs it.
Does blocking ClaudeBot also block Claude for a user?
No. They are separate tokens. Add a Disallow for each one you want to stop.
Does it follow robots.txt?
Yes. Anthropic says its bots honor robots.txt, and this token controls which sites user requests can reach.
Is the live reads report free?
Yes. Live reads, AI crawlers and AI referral visits are all in the free plugin. Download the free plugin to try it.