effect-ts

name

effect-ts

description

Effect-TS library usage in TypeScript — Effect.gen generators, Schema.Struct/Schema.Class definitions, Layer/Context.Tag/Service patterns, Effect.pipe pipelines, Data.TaggedError/Data.Class error types, Ref/Queue/PubSub/Deferred concurrency primitives, Match module, Config providers, Scope/Exit/Cause/Runtime patterns, or any code using Effect's typed error channel (E parameter). Trigger when writing, reviewing, debugging, or refactoring TypeScript code that uses Effect — when you see imports from `effect`, `effect/*`, or any `@effect/*` scoped package (schema, platform, sql, opentelemetry, cli, cluster, rpc, vitest). Also trigger when the user asks about Effect patterns, migration from Promises/fp-ts/neverthrow to Effect, or how to structure an Effect application. Do NOT trigger for React's useEffect, Redux side effects, or general English usage of "effect" unless the context clearly involves the Effect-TS library.

Effect TypeScript Best Practices

Effect is a TypeScript library for building complex, type-safe applications with structured error handling, dependency injection via services/layers, fiber-based concurrency, and resource safety.

When to Apply

• Writing or reviewing TypeScript code that imports from effect, @effect/schema, or @effect/platform
• Implementing typed error handling with Effect<Success, Error, Requirements>
• Building services and layers for dependency injection
• Working with Schema for data validation, decoding, and transformation
• Using fiber-based concurrency (queues, semaphores, PubSub, deferred)
• Processing data with Stream and Sink
• Migrating from Promises, fp-ts, neverthrow, or ZIO to Effect

How to Use

This skill is organized by domain. Read the relevant reference file for the area you're working in.

Read First: The Paradigm

Always read this before diving into API references, especially when refactoring existing code to use Effect or writing new Effect services:

Reference	When to Read
Think in Effect: The Paradigm Shift	Before any other reference. Mental model shifts, refactoring recipes, anti-patterns, application architecture. Read this to understand HOW to think in Effect — the other files teach WHAT to type.

Core Foundations

Reference	When to Read
Getting Started	Creating the Effect type, pipelines, generators, running effects
Error Management	Typed errors, recovery, retrying, timeouts, sandboxing
Core Concepts	Request batching, configuration management, runtime system

Data & Validation

Reference	When to Read
Data Types	Option, Either, Cause, Chunk, DateTime, Duration, Exit, Data
Schema Basics	Schema intro, basic usage, classes, constructors, effect data types
Schema Advanced	Transformations, filters, annotations, error formatting, JSON Schema output

Architecture & Dependencies

Reference	When to Read
Requirements Management	Services, Layers, dependency injection, layer memoization
Resource Management	Scope, safe resource acquisition/release, caching
State Management	Ref, SubscriptionRef, SynchronizedRef for concurrent state

Concurrency & Streaming

Reference	When to Read
Concurrency	Fibers, Deferred, Latch, PubSub, Queue, Semaphore
Streams and Sinks	Creating, consuming, transforming streams; sink operations
Scheduling	Built-in schedules, cron, combinators, repetition

Platform & Observability

Reference	When to Read
Platform	FileSystem, Command, Terminal, KeyValueStore, Path
Observability	Logging, metrics, tracing, Supervisor
Testing	TestClock for time simulation; for service mocking and layer testing, see Requirements Management

Style, AI & Migration

Reference	When to Read
Code Style	Branded types, pattern matching, dual APIs, guidelines, traits
AI Integration	Effect AI packages for LLM tool use and execution planning
Micro	Lightweight Effect alternative for smaller bundles
Migration Guides	Coming from Promises, fp-ts, neverthrow, or ZIO

Quick Reference — Common Patterns

The Effect Type

//         ┌─── Success type
//         │        ┌─── Error type
//         │        │      ┌─── Required dependencies
//         ▼        ▼      ▼
Effect<Success, Error, Requirements>

Creating Effects

import { Effect } from "effect"

// From sync values
const succeed = Effect.succeed(42)
const fail = Effect.fail(new Error("oops"))

// From sync code that may throw
const sync = Effect.try(() => JSON.parse(data))

// From promises
const async = Effect.tryPromise(() => fetch(url))

// From generators (recommended for complex flows)
const program = Effect.gen(function* () {
  const user = yield* getUser(id)
  const todos = yield* getTodos(user.id)
  return { user, todos }
})

Running Effects

// Async (returns Promise)
Effect.runPromise(program)

// With full Exit information
Effect.runPromiseExit(program)

// Sync (throws on async)
Effect.runSync(program)

Typed Errors

import { Data, Effect } from "effect"

class NotFound extends Data.TaggedError("NotFound")<{
  readonly id: string
}> {}

class Unauthorized extends Data.TaggedError("Unauthorized")<{}> {}

// Error type is tracked: Effect<User, NotFound | Unauthorized>
const getUser = (id: string) =>
  Effect.gen(function* () {
    // ...
  })

Services and Layers

import { Context, Effect, Layer } from "effect"

// Define a service
class UserRepo extends Context.Tag("UserRepo")<
  UserRepo,
  { readonly findById: (id: string) => Effect.Effect<User, NotFound> }
>() {}

// Use in effects — adds to Requirements channel
const program = Effect.gen(function* () {
  const repo = yield* UserRepo
  return yield* repo.findById("1")
})

// Implement with a Layer
const UserRepoLive = Layer.succeed(UserRepo, {
  findById: (id) => Effect.succeed({ id, name: "Alice" })
})

// Provide and run
program.pipe(Effect.provide(UserRepoLive), Effect.runPromise)

Schema Validation

import { Schema } from "effect"

const User = Schema.Struct({
  id: Schema.Number,
  name: Schema.String,
  email: Schema.String.pipe(Schema.pattern(/@/))
})

type User = typeof User.Type

// Decode (parse + validate)
const decode = Schema.decodeUnknownSync(User)
const user = decode({ id: 1, name: "Alice", email: "[e-mail]" })

Pipelines

import { Effect, pipe } from "effect"

// Data-last (pipe style)
const result = pipe(
  getTodos,
  Effect.map((todos) => todos.filter((t) => !t.done)),
  Effect.flatMap((active) => sendNotification(active.length)),
  Effect.catchTag("NetworkError", () => Effect.succeed("offline"))
)

// Fluent (method style)
const result2 = getTodos.pipe(
  Effect.map((todos) => todos.filter((t) => !t.done)),
  Effect.flatMap((active) => sendNotification(active.length))
)

Gotchas

See gotchas.md for known failure points.