Scenario Trigger Engine

The Scenario Trigger Engine is the highest-complexity component in the notification system, running as a Supabase Edge Function on both database event hooks and periodic cron schedules. It has three upstream dependencies — notification-preferences-repository, scenario-notification-repository, and push-notification-dispatcher — all of which must be stable before end-to-end testing is possible. The two sub-components (Inactivity Detector and Milestone Detector) carry medium complexity each and should be scoped as separate development tasks within the same sprint. Key delivery risks include cron schedule reliability in the Supabase Edge Function environment, cooldown logic correctness to prevent notification spam, and opt-out preference lookups adding latency to the evaluation loop.

Load testing is required for the scheduled evaluation path to ensure it completes within the cron window for large user populations.

The Scenario Trigger Engine is a high-complexity Supabase Edge Function implementing three scenario evaluators against five data models: activity, certification, achievement, notification_preference, and push_notification. Core interfaces include: `evaluateInactivityScenario(userId, thresholdDays)` delegating to the Inactivity Detector sub-component; `evaluateMilestoneScenario(userId, milestoneType)` delegating to the Milestone Detector sub-component; `evaluateCertificationExpiryScenario(userId, leadTimeDays)` querying the certification model directly; `shouldNotify(userId, scenarioType)` checking opt-out preferences and cooldown expiry before any enqueue; `enqueueNotification(userId, scenarioType, contextData)` writing to scenario-notification-repository and calling push-notification-dispatcher; `recordLastTriggered(userId, scenarioType)` updating cooldown state; `runScheduledEvaluation()` iterating all active users on cron; and `handleDatabaseEvent(event)` responding to Supabase realtime database hooks. Cooldown state must be persisted in the database, not in-memory, to survive function cold starts. Each evaluator should be independently testable in isolation.