Where is the Claude Code in-app browser available?

The Claude Code in-app browser ships only in the Desktop app on macOS, Windows, and Linux. The terminal CLI does not have it. Inside the desktop app, Claude opens external sites in a sandboxed webview, safety classifiers review each action on the page, and users choose whether browsing sessions persist between runs or start fresh every time.

What can /doctor do in Claude Code Week 28 that it could not do before?

In Week 28, /doctor stops being a read-only report and becomes a repair tool. It checks install health, flags unused skills, MCP servers, and plugins that cost more context than they earn, deduplicates local CLAUDE.md files against checked-in copies, proposes trimming CLAUDE.md content Claude could derive from the codebase, and reports slow hooks. It always confirms before making any change.

Which point releases are covered in the Claude Code Week 28 digest?

The Week 28 digest covers Claude Code v2.1.202 through v2.1.206, shipped between July 6 and 9, 2026. Highlights include the /doctor rebuild in v2.1.205, an auto mode rule in v2.1.205 that blocks tampering with session transcript files, directory suggestions on /cd in v2.1.206, and a Claude Opus 4.8 quality bump for the /code-review command.

How does the desktop browser change how Claude Code fetches docs?

Before Week 28, the Claude Code Desktop app could preview a local dev server, but external documentation and design tools had to be pasted in by hand. With the in-app browser, Claude opens the same sites directly, reads them like a page, and can click through, fill fields, and interact the same way it does with a preview, while the safety classifier gates risky actions.

Anthropic · 2026-07-10 · major

Claude Code Desktop — in-app browser lands and /doctor gains a repair mode

Claude Code Week 28 ships a built-in browser in the desktop app, upgrades /doctor from a read-only report to a fixer, and adds an auto mode rule that blocks tampering with session transcripts.

Claude Code Week 28 gives the desktop app a real browser and turns /doctor from a report into a fixer.

Key specs

Versions	v2.1.202–v2.1.206
Week	28 (Jul 6–10, 2026)

Quick facts

Product	Claude Code
Maker	Anthropic
Versions	v2.1.202 – v2.1.206 (Jul 6–10, 2026)
Headline feature	In-app browser in the desktop app
Second feature	/doctor rebuilt as a full setup checkup
Auto mode	Now blocks tampering with session transcripts
Availability	Claude Code Desktop on macOS, Windows, and Linux

What is it?

The Week 28 digest bundles two headline features for Claude Code. The desktop app now has a built-in browser: the coding agent can open any external site, read the page, click through links, and interact with forms and buttons the same way it already worked with a local dev-server preview. The browser is sandboxed, safety classifiers review every action on external pages, and users pick whether browsing sessions persist between runs.

How does it work?

The in-app browser is a sandboxed webview embedded in the Claude Code Desktop shell, sitting alongside the existing dev-server preview panel. When Claude clicks or types on an external site, a safety classifier runs first and can refuse. The rebuilt /doctor command runs a battery of checks — install health, unused skills, MCP servers, and plugins measured against their context cost, duplicate CLAUDE.md files, and slow hooks — then asks the user before applying each fix. Its alias is /checkup.

Why does it matter?

The desktop browser removes a common Claude Code workaround where developers had to paste external documentation into the terminal by hand or ask Claude to open a URL through a shell command. A working /doctor turns install and configuration debugging into a one-command loop instead of a scavenger hunt through logs, and Week 28's auto mode rule that blocks tampering with transcript files closes a known way agentic mode could rewrite its own audit trail.

Who is it for?

Claude Code Desktop users

Frequently asked questions

Where is the Claude Code in-app browser available?: The Claude Code in-app browser ships only in the Desktop app on macOS, Windows, and Linux. The terminal CLI does not have it. Inside the desktop app, Claude opens external sites in a sandboxed webview, safety classifiers review each action on the page, and users choose whether browsing sessions persist between runs or start fresh every time.
What can /doctor do in Claude Code Week 28 that it could not do before?: In Week 28, /doctor stops being a read-only report and becomes a repair tool. It checks install health, flags unused skills, MCP servers, and plugins that cost more context than they earn, deduplicates local CLAUDE.md files against checked-in copies, proposes trimming CLAUDE.md content Claude could derive from the codebase, and reports slow hooks. It always confirms before making any change.
Which point releases are covered in the Claude Code Week 28 digest?: The Week 28 digest covers Claude Code v2.1.202 through v2.1.206, shipped between July 6 and 9, 2026. Highlights include the /doctor rebuild in v2.1.205, an auto mode rule in v2.1.205 that blocks tampering with session transcript files, directory suggestions on /cd in v2.1.206, and a Claude Opus 4.8 quality bump for the /code-review command.
How does the desktop browser change how Claude Code fetches docs?: Before Week 28, the Claude Code Desktop app could preview a local dev server, but external documentation and design tools had to be pasted in by hand. With the in-app browser, Claude opens the same sites directly, reads them like a page, and can click through, fill fields, and interact the same way it does with a preview, while the safety classifier gates risky actions.

Try it

Update Claude Code, then run /doctor from any session