Agents used to live inside a modal — click edit, get a cramped popup, close it, lose your place. Today every agent gets a real home: /app/agents/[id], a dedicated page that replaces the old edit modal entirely.
One page, four tabs
The workspace page is organised into four tabs, and each one covers a distinct part of what makes an agent an agent.
Context @researcher
@researcher what did we learn about pricing last week?
Three takeaways from the competitor sweep: usage-based beats seats in this segment, the $20 anchor holds, and annual discounts convert best at 2 months free.
@researcher 3 memories used
Attach an agent with @ — its memory and apps come with it.
Config is the agent's identity: its name, its instructions, the model it prefers, any apps it has pinned, sub-agents it can call on, its memory settings, and — new alongside the artifacts library — exactly which artifacts and folders it's been granted, and at what access level. Everything you used to set through the modal, now with room to actually see what you're doing.
Memory opens onto the agent's private partition — the facts it holds and what it's learned on its own, separate from your account memory. It was always inspectable; now it's a first-class tab instead of a secondary screen you had to know to find.
Skills shows what the agent is equipped to do beyond its base instructions — the specific capabilities layered on top of a name and a prompt.
Logs is the agent's own activity feed — a record of what it's done, scoped to that agent specifically, rather than mixed into your general activity. If you're trying to figure out why an agent behaved a certain way in a chat last week, this is where you go looking, not through a shared history that buries it among everything else.
Why a page instead of a modal
Modals are fine for quick edits. They're bad for anything you actually want to sit with — comparing what an agent remembers against what it's configured to do, or checking its recent activity while you tweak its instructions. A dedicated page means all of that is one tab-click away instead of a stack of things you had to open, close, and reopen to cross-reference.
It also means an agent's page is a real, shareable location — a URL you can return to, not a transient overlay tied to wherever you happened to click edit from. Bookmark it, jump straight to an agent's Logs tab after a run finishes, or send a teammate directly to its Config tab instead of describing which buttons to click.
The timing isn't a coincidence, either. Now that agents can hold real access to your artifacts library — specific files or whole folders, at read or read-write — there's meaningfully more to configure per agent than there used to be. A modal was already tight before that; it wasn't going to survive a permissions list on top of everything else. The Config tab has room for it now, laid out instead of squeezed in.
The old modal is gone. Open any agent from the Agents page and you land here now — everything about that specialist, in one place.
Try it yourself
Every model, one memory, and agents that work the way you do.