andrueandersoncs/claude-skill-effect-ts

Overview

This skill explains how Effect optimizes API calls through batching, caching, and deduplication. It shows how to define request types and resolvers, enable automatic batching, and use built-in caching utilities like Effect.cached and the Cache service. The goal is to eliminate N+1 request patterns and reduce redundant network or database work.

How this skill works

Effect inspects request operations and can aggregate many individual requests into a single batched resolver call, allowing one network/database call to satisfy multiple logical requests. It also memoizes effect results and request outcomes within a query context, preventing duplicate concurrent work and enabling TTL-based or manual invalidation. You can opt in or out of batching and provide custom caches or resolvers with access to contextual services.

When to use it

• When facing N+1 query problems across loops or forEach operations
• When multiple callers request the same resource concurrently and you want deduplication
• When API or DB supports batch endpoints to reduce round trips
• When you need short-term memoization of expensive computations or network calls
• When you want fine-grained control over cache TTL, capacity, or invalidation

Best practices

• Use batched RequestResolvers at API boundaries to group related IDs into a single call
• Prefer Effect.cached or cachedWithTTL for inexpensive, read-heavy values to reduce repeated fetches
• Set TTLs to balance freshness and performance; use invalidate hooks for known changes
• Rely on automatic request deduplication within a query context to avoid duplicate inflight work
• Disable batching or request caching selectively for operations that must run independently

Example use cases

• Batching user lookups when iterating todos to avoid N+1 requests to /api/users
• Using RequestResolver.makeBatched to POST an array of ids and distribute results to individual requests
• Caching application config with Effect.cached so subsequent reads are instantaneous
• CachedWithTTL for current user data with automatic expiry after a short interval
• Creating a Cache service to control capacity, TTL, manual invalidation, and statistics