Explanation¶

Why the app works the way it does: the ideas behind it and the trade-offs it makes. Nothing here needs doing; it is for understanding.

• Why Agents: why you would run your coding agents here, alongside the editor you already use, and when something else is better.
• The control plane: hosts, clients, copies and the store, where to run it, and what happens when it is down.
• The window and the host: why agents keep working when the window is closed, what Agents Host does, and what happens when the Mac restarts.
• Why agents' own tools are taken away: what a runtime loses in the app's sessions, what it keeps, and why your own setup is untouched.
• Views in a conversation: how a tool's ui:// view is drawn on every device, what it may reach, and what its messages do.
• Projects, hosts and worktrees: a project is a folder on one machine, and an agent can have a worktree of its own.
• Leases on shared resources: how agents take turns with the screen, a browser or a simulator, and what you can end.
• How the phone and iPad reach your agents: every host, at home and away, grants, and forgetting a device.
• Compaction in each runtime: whether a long conversation compacts itself in each runtime, what the app sees when it does, and what it could pass.
• Token-compression tools: whether rtk or Headroom would cut what agents cost, how either would plug in, and what the app could do instead.
• What agents are sent each turn: what the app's own tools add to every request, what is cached, and what was changed to cut it.
• Direction: what the project is working towards now, what it is not, and how the work is checked against it every few days.