Can Claude search the web?

Yes. Anthropic's help article on web search says that when search is on, every response includes citations so you can check the sources. Responses can include direct citations, source links for further reading, and relevant quotes.

• Turn it on. In the chat window, select the plus button and choose Web search. On the newer Claude experience there is no toggle. The article says Claude searches the web when it helps.
• Team and Enterprise plans. An Owner or Primary Owner must first enable web search for the workspace under Organization settings, Capabilities. Members can then turn it on in a chat.
• Free accounts. Usage limits apply, and web search counts toward them. The limits reset every five hours.
• Location. Claude may use a location inferred from your IP address when you ask for localized results.

If web search is missing for you, check your plan, the workspace setting and your usage limit first. These are the settings the help article describes.

Developers can give Claude a web search tool in the Messages API. Anthropic's documentation says Claude decides when to search based on the prompt, the API runs the searches, and Claude answers with cited sources. Claude searches when a request depends on information that is current, changing or outside its training data, such as recent events, current prices, or facts about organizations that may have changed. Claude answers directly when the request draws on stable knowledge.

A separate web fetch tool retrieves the full content of specific pages and PDFs. Anthropic documents three limits that matter to site owners:

• Claude can only fetch URLs that already appeared in the conversation. These are URLs in user messages, in tool results, or in earlier search and fetch results. Search results are the usual source of new URLs.
• The fetch tool does not support pages that depend on JavaScript rendering. Anthropic's documentation says so directly. Content that exists only after scripts run is not available to it.
• The API can refuse a fetch. The error list names Anthropic side restrictions such as private addresses and robots.txt among the reasons a URL can be refused.

Organization admins can turn web search and web fetch off in the Claude Console, and can set allowed or blocked domains. These controls belong to the developer who uses the API.

Yes. Claude Code documents two tools. WebSearch runs a query against Anthropic's web search backend and returns result titles and URLs. It does not fetch the result pages. WebFetch takes a URL and a prompt, fetches the page, and converts HTML to Markdown. To read a page found in search results, Claude follows up with WebFetch.

Claude Code's documentation says WebFetch sets a User-Agent header that begins with Claude-User. It also sends an Accept header that prefers Markdown over HTML, so servers that support content negotiation can return Markdown. In your logs, a request from Claude Code's WebFetch therefore appears as Claude-User. See the Markdown for agents fix.

Claude Code's documentation says WebFetch asks the user for permission before it fetches a domain, unless a permission rule or a built in set of preapproved documentation domains covers it.

Anthropic's help article on web search and its API documentation checked on 2026-11-06 do not name a provider for general web search. The help article does say that image results are powered by Bing.

In March 2025, press reports said Anthropic had added Brave Search to its subprocessor list on 2025-03-19. TechCrunch (2025-03-21) reported that engineer Antonio Zugaldia spotted the change and that Anthropic had not commented. Developer Simon Willison ran a test search that returned the same citations in Claude and in Brave Search, and he reported a Brave related parameter in Claude's search function definition. These are press reports and one independent test.

SeenLens could not confirm the 2025 listing against Anthropic's own subprocessor page for 2026. Anthropic's trust center page returned no list to our fetch tool. The Brave Search connection is reported for 2025 and is unconfirmed for 2026. Anthropic does not say that Brave Search indexing changes your visibility in Claude.

Anthropic says its bots honor industry standard robots.txt directives, and that it supports the non-standard Crawl-delay extension. The article gives no user agent strings. Match on the bot names.

Use robots.txt to set each bot separately. This example allows Claude's search and user bots and opts out of training:

User-agent: Claude-SearchBot
Allow: /

User-agent: Claude-User
Allow: /

User-agent: ClaudeBot
Disallow: /

• Do not block by IP address alone. Anthropic says IP blocking may not work correctly or persistently guarantee an opt-out, because it stops Anthropic from reading robots.txt. Anthropic publishes its IP ranges at claude.com/crawling/bots.json.
• Serve content as HTML. The API web fetch tool does not support JavaScript rendered pages. See content that needs JavaScript.
• Check your firewall. A firewall rule can block these bots while robots.txt allows them. See AI crawlers blocked by a firewall.
• Training preference signal. Anthropic says that from 2026-10-30 it honors a machine readable signal for a no AI training preference. The rollout is ongoing, the signal is based on an IETF draft, and handling may change. A robots.txt rule that names ClaudeBot takes priority over the general setting.

The robots.txt guide for AI crawlers covers all AI bots, and the Content Signals guide explains the preference lines.

1. Run the AI crawler checker above. It reads robots.txt for ClaudeBot, Claude-SearchBot and Claude-User. It also requests your page with the ClaudeBot user agent and compares the response with a browser request.
2. Search your logs for Claude-SearchBot, Claude-User and ClaudeBot. Compare source addresses with Anthropic's published IP list.
3. Ask Claude with web search on a question that your page answers, and look at the citations. One answer is a spot check and not proof.

Check your site with the free SeenLens scan

• Anthropic Help Center: Enabling and using web search (plans, how to enable, citations, image results powered by Bing, usage limits; checked 2026-11-06)
• Anthropic Help Center: Does Anthropic crawl data from the web, and how can site owners block the crawler? (ClaudeBot, Claude-SearchBot, Claude-User, robots.txt, Crawl-delay, IP list, no AI training signal; checked 2026-11-06)
• Claude API docs: Web search tool (when Claude searches, citations, admin controls; checked 2026-11-06)
• Claude API docs: Web fetch tool (URL validation, no JavaScript rendering, robots.txt among refusal reasons; checked 2026-11-06)
• Claude Code docs: Tools reference (WebSearch, WebFetch, Claude-User user agent, Accept header; checked 2026-11-06)
• TechCrunch: Anthropic appears to be using Brave to power web searches for its Claude chatbot (press report, 2025-03-21)
• Simon Willison's Weblog: Anthropic and Brave Search (independent test, 2025-03-21)