effect core

📁 andrueandersoncs/claude-skill-effect-ts 📅 Jan 1, 1970
8
总安装量
0
周安装量
#34188
全站排名
安装命令
npx skills add https://github.com/andrueandersoncs/claude-skill-effect-ts --skill 'Effect Core'

Skill 文档

Effect Core

CRITICAL: You MUST Follow the Code Style Skill

Before writing ANY Effect code, you MUST read and follow the Code Style skill (view on GitHub). This skill covers core APIs and composition only — the Code Style skill defines the mandatory patterns, forbidden anti-patterns, and idiomatic conventions that apply to ALL Effect code you produce. Every code example you generate must conform to those rules.

Overview

Effect is the foundational type in Effect-TS representing a computation that may succeed with value A, fail with error E, or require context R:

Effect<Success, Error, Requirements>;
// Also written as: Effect<A, E, R>

The key insight: Effects are descriptions of programs, not executed code. They must be explicitly run.

Creating Effects

From Synchronous Values

import { Effect } from "effect";

const success = Effect.succeed(42);

const failure = Effect.fail(new Error("Something went wrong"));

const lazy = Effect.sync(() => {
  console.log("Computing...");
  return Math.random();
});

// Sync that may throw (converts exception to typed error)
const mayThrow = Effect.try({
  try: () => someLegacyFunction(),
  catch: (error) => new LegacyError({ cause: error }),
});

// For JSON parsing, prefer Schema.parseJson (type-safe and validated)
const UserInput = Schema.parseJson(
  Schema.Struct({
    name: Schema.String,
    value: Schema.Number,
  }),
);
const parsed = Schema.decodeUnknown(UserInput)(userInput);

From Asynchronous Values

// From Promise (untyped error)
const fromPromise = Effect.promise(() => fetch("/api/data"));

// From Promise with typed error
const fromPromiseTyped = Effect.tryPromise({
  try: () => fetch("/api/data"),
  catch: (error) => new FetchError({ cause: error }),
});

Composing Effects

Sequential Composition with pipe

import { Effect, pipe } from "effect";

const program = pipe(
  Effect.succeed(1),
  Effect.map((n) => n + 1),
  Effect.flatMap((n) => Effect.succeed(n * 2)),
  Effect.andThen((n) => Effect.succeed(`Result: ${n}`)),
);

Generator Syntax (Recommended)

The generator syntax (Effect.gen) provides cleaner, more readable code:

const program = Effect.gen(function* () {
  const a = yield* Effect.succeed(1);
  const b = yield* Effect.succeed(2);
  const result = a + b;
  return `Sum: ${result}`;
});

Equivalent to:

const program = Effect.succeed(1).pipe(
  Effect.flatMap((a) => Effect.succeed(2).pipe(Effect.flatMap((b) => Effect.succeed(`Sum: ${a + b}`)))),
);

Running Effects

Effects are descriptions that must be run to produce values:

import { Effect } from "effect";

const program = Effect.succeed(42);

const result = await Effect.runPromise(program);

const syncResult = Effect.runSync(Effect.succeed(42));

const exit = await Effect.runPromiseExit(program);

Runtime Methods

Method Use Case
Effect.runPromise Async effect → Promise (throws on failure)
Effect.runPromiseExit Async effect → Promise (never throws)
Effect.runSync Sync effect → value (throws on async/failure)
Effect.runSyncExit Sync effect → Exit (throws on async)

Key Composition Operators

map – Transform Success Value

Effect.succeed(5).pipe(Effect.map((n) => n * 2));

flatMap / andThen – Chain Effects

const getUser = (id: number) => Effect.succeed({ id, name: "Alice" });
const getPosts = (userId: number) => Effect.succeed([{ title: "Post 1" }]);

const program = getUser(1).pipe(Effect.flatMap((user) => getPosts(user.id)));

tap – Side Effects Without Changing Value

Effect.succeed(42).pipe(
  Effect.tap((n) => Effect.log(`Got value: ${n}`)),
  Effect.map((n) => n * 2),
);

all – Combine Multiple Effects

// Tuple of effects
const tuple = Effect.all([Effect.succeed(1), Effect.succeed("hello"), Effect.succeed(true)]); // Effect<[number, string, boolean], never, never>

// Object of effects
const obj = Effect.all({
  id: Effect.succeed(1),
  name: Effect.succeed("Alice"),
}); // Effect<{ id: number; name: string }, never, never>

Effect vs Promise Comparison

Promise Effect
new Promise((resolve) => resolve(1)) Effect.succeed(1)
Promise.reject(error) Effect.fail(error)
promise.then(f) effect.pipe(Effect.map(f))
promise.then(f) (f returns Promise) effect.pipe(Effect.flatMap(f))
Promise.all([...]) Effect.all([...])
await promise yield* effect (in Effect.gen)

Dual APIs

Most Effect functions support both “data-first” and “data-last” (pipeable) styles:

// Data-last (pipeable) - recommended
Effect.succeed(1).pipe(Effect.map((n) => n + 1));

// Data-first
Effect.map(Effect.succeed(1), (n) => n + 1);

Best Practices

Do

  1. Use Effect.gen for sequential code – More readable than nested flatMaps
  2. Use typed errors – Always define error types with Schema.TaggedError
  3. Use Schema.parseJson for JSON – Never use raw JSON.parse()
  4. Prefer data-last (pipeable) – Consistent with Effect ecosystem

Don’t

  1. Don’t mix async/await with Effect – Use Effect.promise at boundaries only
  2. Don’t use try/catch – Use Effect.try or Effect.tryPromise
  3. Don’t throw exceptions – Use Effect.fail with typed errors
  4. Don’t use JSON.parse – Use Schema.parseJson with a schema

For the complete list of mandatory patterns, forbidden anti-patterns, and idiomatic conventions, see the Code Style skill (GitHub).

Additional Resources

For comprehensive documentation on all Effect APIs, patterns, and advanced usage, consult the full Effect documentation at ${CLAUDE_PLUGIN_ROOT}/references/llms-full.txt.

Search for these sections:

  • “Effect vs Promise” for migration patterns
  • “Getting Started” for installation and setup
  • “Basic Concurrency” for parallel execution
  • “Dual APIs” for function calling conventions