Agentic Apps · 04.09
Live Ops & the App Shell
Navigation, status, approvals, refresh, and an ops drawer that always resolves live active runs — how every App surface stays operable.
A phantom 'running' state is structurally impossible
Every App surface is wrapped by a common shell providing navigation, status, pending approvals, refresh, and an operations drawer that surfaces active runs. That drawer doesn't cache a stale "is something running" flag — it resolves live active runs from the actual run registry every time, which is what makes a phantom "still running" indicator (a UI stuck showing activity for a run that already finished) structurally impossible rather than merely rare.
Readiness has four visible facets
The App compiler reports Structure, Execute, Verify, and Deliver separately in the shell. Structural and executable blockers decide whether an operator can begin a proof run; verification and delivery evidence decide whether its result is ready to trust or release. Runtime, channel, surface, and test checks feed those facets, so a beautiful interface cannot conceal a missing runtime, permission mismatch, or unverified business outcome.
Continue
Live co-presence over the realtime bus, ephemeral by design, so two operators can tell they're looking at the same App at once.
A real project-management backlog for agent work: status, priority, labels, recurrence, and assignment.