What problem does it solve? Mobile teams shipping Expo apps lack visibility into production startup and navigation performance, making it hard to detect slow cold launches, blocked main threads, or network-bound startups before users complain. ## Core Features & Use Cases - Instrumentation Setup: Add expo-observe to an Expo project with the correct SDK 55 vs SDK 56+ APIs (AppMetricsRoot/ObserveRoot, markInteractive, useObserve, ObserveInteractiveMarker), plus Expo Router and React Navigation integrations for per-route metrics. - CLI Querying: Run the six eas observe:* commands (metrics-summary, metrics, routes, events, session, versions) with correct flags, aliases, pagination, and JSON output shapes. - Metrics Interpretation: Diagnose slow startups using TTI frame-rate, device, and network params to distinguish main-thread contention, hard freezes, and throttled devices. - Use Case: After a release, run eas observe:metrics tti --sort slowest --days 7, grab a slow sample's sessionId, and replay the full session timeline with eas observe:session to find the blocking request. ## Quick Start Use the eas-observe skill to add expo-observe instrumentation to my Expo SDK 56 project and show me how to query last week's slowest TTI samples.