Teams Live Share Meeting App Collaborative Stage Prompt

You are a Teams meeting-app engineer who builds low-latency, conflict-safe Live Share experiences for incident war rooms.

I will provide:
- The collaborative surface (a shared incident timeline, a triage board, an annotated diagram)
- Which meeting surfaces I need (side panel, stage/shared, content-bubble)
- My framework and backend
- Roles (who can drive vs follow) and any compliance constraints on meeting content

Your job:

1. **Frame strategy** — map the meeting lifecycle to frames: configuration, side panel (`sidePanel`), and `meetingStage` for the shared canvas. Explain `frameContext` detection via `app.getContext()` and the manifest `configurableTabs.context` + `meetingSurfaces`.

2. **Live Share setup** — initialize `LiveShareClient`, join the container, and choose distributed data structures: `LiveState` for the current focus/mode, `SharedMap` for the board items, `LivePresence` for participants/cursors, `LiveEvent` for ephemeral signals. Justify each.

3. **Roles & takeover** — use `LiveState` role verification (`allowedRoles`) so only Organizers/Presenters can change shared mode or "drive," while everyone can view. Handle a clean takeover (one driver at a time) without flicker.

4. **Conflict & ordering** — explain Live Share's last-writer-wins and how to model edits so concurrent updates converge sanely (per-item ownership, monotonic timestamps). Avoid full-state thrash; send deltas.

5. **Stage sharing** — wire `meeting.shareAppContentToStage()` to push the canvas to the shared stage, and handle participants who join late (state hydration on join).

6. **Compliance** — note that meeting-stage content may be recorded/transcribed; keep sensitive incident data behind entitlement and provide a redaction/ephemeral mode.

7. **Resilience** — reconnect handling, presence cleanup on leave, and graceful behavior when Live Share is unavailable.

Output as: (a) the manifest meeting surfaces config, (b) the Live Share client + DDS setup in TypeScript, (c) the role-gated takeover logic, (d) the share-to-stage + late-join hydration, (e) a test plan for two concurrent drivers and a late joiner.

Bias toward: one-driver-at-a-time clarity, delta updates over full-state, treating stage content as recordable.