Anthropic has integrated a browser into Claude Code, enabling the AI to interact with external websites by reading, clicking, and typing. This feature allows users to access documentation sites, issue trackers, and similar resources directly within the app. The browser functions like a tab-based interface and can be opened with a keyboard shortcut. According to the company, the browser uses the same tools as the app’s local app preview but includes additional safety measures.
Claude Code’s browser includes classifiers that screen any write actions on external sites, ensuring the AI does not perform actions like purchasing items, creating accounts, or bypassing CAPTCHAs without user consent. The browser operates on a clean profile with no saved logins, and organizations can restrict access to external sites through an allowlist or disable the browser tools entirely. Anthropic announced these updates on X, emphasizing the feature’s controlled environment for web interaction.
Claude Code’s browser is designed to complement the Chrome extension for users who want the AI to act within their own logged-in sessions. The feature aims to enhance productivity by allowing AI to engage with web content while maintaining security and user control. Source: thedecoder
