Release 1.50.0

One panel, the whole screen

Every panel header now has a maximize button, next to minimize and its opposite: instead of tucking this panel away, it tucks away all the others and lets the one that's left fill the entire app.

It's how you read a browser, follow a terminal or look at a design with nothing around it. Click again and everyone comes back exactly where they were, including what you had already tucked away yourself, which stays tucked away.

Each work tab has its own, so maximizing in one place doesn't undo the layout in another.

Sign in once, and every browser in the project comes along

Until now, the browser you opened and the one an agent opened were separate worlds. That protects what's yours: an agent never reaches the login you made in your own window. But it costs a lot for anyone showing work to a client, because every preview someone opens starts back at the sign-in screen.

Now it's the project's choice. Under Edit canvas → Browser, the Browser session setting accepts "Shared in this project": your browser and the ones belonging to that canvas's agents start sharing the same logins. Sign in once, and it holds for every preview anyone opens, today and tomorrow.

What gets shared stays right there. Each project has its own, none can see another's, and whatever you have signed in on your own browsers lives somewhere else, handed to no one you didn't tell.

Separate sessions remain the default.

A browser that knows no one

Right-click any browser panel and you'll find Clean session: it forgets everything and browses as a visitor. Good for testing a sign-up from scratch, or seeing the site the way a first-time visitor sees it. Next to it, Sign out of every session in this project.

And you can just ask an agent: "open the preview signed out".

A screenshot from your phone, straight to the agent

In Canvas Companion, the iPhone app, the message field now has a paperclip. Take a photo or pick one from your library, type your question alongside it if you want, and send.

On the Mac the image lands on the canvas next to the agent's panel, and its path goes straight into the prompt line, just like dragging a photo onto an agent. Send only the image and the path waits in the agent's box, so you can write the question whenever you like, from your phone or your desk.