Anthropic launches in-app browser for Claude Code to streamline developers' workflows

2 Sources

Share

Anthropic has introduced an in-app browser for Claude Code on desktop, allowing the AI coding assistant to open websites, read documentation, and interact with web pages without switching windows. The sandboxed environment includes security safeguards and site-specific permissions, while keeping browsing isolated from personal profiles to help developers focus on building and testing applications.

News article

Anthropic Brings Native Web Browsing to Claude Code Desktop

Anthropic has equipped Claude Code with an in-app browser that fundamentally changes how developers interact with their AI coding assistant

1

. The desktop app now allows Claude Code to browse the web, open documentation, inspect designs, and interact with web apps directly within the application itself. Developers can launch the browser using Ctrl + Shift + B on Windows or Cmd + Shift + B on macOS, eliminating the need to constantly switch between code editors and browser tabs

2

.

How the AI Coding Assistant Navigates Web Content

The Browser pane functions as a fully tabbed browser that sits alongside a developer's workspace, capable of reading page contents, clicking links, and interacting with elements much like it already does with local development environments

2

. External links shared inside Claude conversations can be opened directly in the Browser pane or in the user's default browser. The browser even supports website logins, including Google OAuth pop-ups, making it practical for building and testing applications that require authentication. This capability addresses one of the biggest annoyances in AI-assisted coding: constantly copying links back and forth between a browser and the coding assistant.

Security Safeguards and Sandboxed Environment Protection

Anthropic has implemented multiple security safeguards to protect users while Claude interacts with live websites. The first time Claude attempts to act on a particular website, users receive a prompt asking whether to Allow once, Always allow, or Deny that action

2

. These site-specific permissions are stored per site and can be revoked from the settings menu at any time. Even after approval, Claude cannot create accounts, make purchases, or bypass CAPTCHAs without explicit user permission. The Browser pane operates in a sandboxed environment with isolated browser profiles, using a clean browser profile separate from personal browsing data, with none of the user's saved logins or history

1

.

Competing with OpenAI and Chrome Extension Alternatives

The announcement comes as agentic web browsing gains momentum, with OpenAI recently revealing plans to sunset its ChatGPT Atlas browser in favor of the new ChatGPT desktop app's in-app browser

1

. Both Anthropic and OpenAI offer Chrome extension options for integrating their AI tooling with Google's browser. Anthropic's documentation clarifies the distinction: the built-in browser is ideal for development and testing, while users who want Claude to work with their existing browser sessions and logged-in accounts should use the Claude in Chrome extension instead, which shares the browser's login state

1

. According to the ClaudeDevs account, sessions are configurable, with users choosing whether they persist

1

. This update positions Anthropic to streamline developers' workflows by consolidating multiple tools into a single interface, potentially reducing context-switching and improving productivity for teams that read documentation and interact with web apps throughout their development cycles.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved