# Supply Chain Rescue

## Platform
Web.

## Stack
- SAP CAP Node.js 10.1 on Node 24, with persistent SQLite.
- SAPUI5 1.136.x (Horizon), loaded from the SAP CDN and served by CAP.
- All app data goes through OData V4.
- The backend needs no model, no LLM and no network.

## Product purpose
Explore a fictional electric motorcycle supply chain:
1. Disrupt a supplier or a plant.
2. See which parts and models are affected, and why.
3. Choose standby backups and preview the recovery the server calculates.
4. Approve, reject or cancel it explicitly.
5. Keep the decision history.

Historical incidents can be browsed with plain filters.

## Capabilities and constraints
- **Network:** 40 nodes: 12 suppliers (3 backups), 20 parts, 3 plants and 5 models.
- **Plan:** an illustrative plan of 320 motorcycles a week.
- **Scenarios:** 4.
- **Incidents:** 24 fictional incidents.
- **Impact:** follows each model's own part requirements, not graph reachability.
- **Simulation limits:** binary availability, immediate recovery, unlimited backup capacity. No forecasting, inventory or prices.
- **Deferred:** agent recommendations, semantic incident matching and notifications (spec A3).

## Users
Portfolio and demo visitors. A presenter signs in as `approver`; `planner` cannot approve.

## Product principles
- Explain each consequence.
- Distinguish unavailable supply from downstream exposure.
- The server owns every calculation and rule; the browser only presents.
- Backups are the user's choice, never a recommendation.
- Say only what is true: no claimed notifications and no implied AI evidence.