navan-performance-tuning

# Navan Performance Tuning ## Overview Navan's REST API has no bulk endpoints or GraphQL — every data fetch is a separate HTTP request. High-volume integrations syncing thousands of bookings, expenses, or user records quickly become bottlenecked by sequential API calls, redundant fetches, and naive pagination. This skill provides concrete optimization patterns: response caching with data-type-specific TTLs, parallel request execution with concurrency controls, cursor-based pagination handling, and HTTP connection reuse. Each pattern targets the real constraint: minimizing total API calls while staying under rate limits. ## Prerequisites - **Active Navan integration** with OAuth 2.0 credentials (client_credentials grant) - **Node.js 18+** (for native fetch and AbortController) - **Understanding of your data volume** — bookings/day, users, expense reports/month - API base URL: `https://api.navan.com/v1` ## Instructions ### Step 1 — Implement Response Caching with Data-Appropriate TTLs Different Navan data types change at different rates. Cache accordingly: ```typescript interface CacheEntry<T> { data: T; expires_at: number; etag?: string; } // TTLs based on data volatility const CACHE_TTL: Record<string, number> = { 'users': 3600_000, // 1 hour — user profiles rarely change 'policies': 86400_000, // 24 hours — travel policies change infrequently 'bookings': 300_000, // 5 minutes — bookings update frequently 'expenses': 600_000, // 10 minutes —...