Why does Muster have a browser, and what is it for?

The built-in browser

Muster has a browser tab. It exists because a large part of verifying front-end work is looking at a page, and because an agent that can look at the page it just changed is a different agent from one that cannot.

What it is

A real browser view in a tab of its own, with a login that persists. The session partition is per project and per CLI profile, so the profile selected in the title bar decides which login the page sees. Signing in once survives restarts.

What the agent can and cannot do

The agent drives the browser through four MCP tools — open a page, read what is on screen, act on it, and report console errors and failed requests.

It navigates on its own only to localhost or to a host you have already allowed. Anything else raises a permission bar, and the bar is kept with the tab even when the tab is in the background. So an agent can open the dev server it just started without asking, and cannot wander onto the open web because a page told it to.

What it is for

• Checking that the page the task changed still renders, and that the console is clean.
• Driving something that has no DOM — a canvas, a map, an editor — because the tools inject real mouse and keyboard events, not synthetic DOM events.
• Reproducing a bug once, in front of you, instead of describing it.

What it is not

It is not a replacement for your browser, and it is not where you browse. It has no history UI, no bookmarks, no profile management. It is an instrument pointed at the thing being built.