Agent skill
effect-advanced
Advanced Effect-TS patterns for typed errors, dependency injection, concurrency, resource management, schema validation, and streaming. Use when building Effect programs — not simple Effect.succeed/fail questions, but multi-concern tasks like designing service layers with Layer composition, handling typed error hierarchies with tagged errors, managing concurrent fibers with structured concurrency, scoped resource lifecycles, schema-driven API contracts, or integrating Effect with existing Express/Hono/database stacks. Do not use for basic TypeScript or general functional programming questions.
Install this agent skill to your Project
npx add-skill https://github.com/trancong12102/agentskills/tree/main/effect-advanced
SKILL.md
Effect Advanced: Patterns, Conventions & Pitfalls
This skill defines the rules, conventions, and architectural decisions for building production Effect-TS applications. It is intentionally opinionated to prevent common pitfalls and enforce patterns that scale.
For detailed API documentation, use other appropriate tools (documentation lookup, web search, etc.) — this skill focuses on how and why to use Effect idiomatically, not the full API surface.
Table of Contents
- Core Conventions
- Error Handling Philosophy
- Dependency Injection Architecture
- Resource & Scope Rules
- Concurrency Model
- Common Pitfalls
- Reference Files
Core Conventions
Use Effect.gen for business logic
Generators read like synchronous code and are strongly preferred over long .pipe /
.flatMap chains for anything beyond trivial composition:
const program = Effect.gen(function* () {
const config = yield* ConfigService;
const user = yield* UserRepo.findById(config.userId);
return user;
});
Reserve pipe for data transformation pipelines and short combinator chains.
Never throw — use Effect's error channel
| Instead of... | Use |
|---|---|
throw new Error() |
Effect.fail(new MyError()) |
try/catch on promises |
Effect.tryPromise({ try, catch }) |
| Callback APIs | Effect.async((resume) => ...) |
| Unrecoverable crashes | Effect.die(defect) |
Functions over methods
Prefer Effect.map(e, f) over e.pipe(Effect.map(f)) for composability and
tree-shaking. Flat imports (import { Effect } from "effect") are fine for
applications; namespace imports (import * as Effect from "effect/Effect") are
better for libraries.
@effect/schema is deprecated
Schema has been merged into core effect. Import from "effect" directly:
import { Schema } from "effect";
// NOT: import { Schema } from "@effect/schema"
Use NodeRuntime.runMain in production
Effect.runPromise does not handle SIGINT/SIGTERM gracefully:
import { NodeRuntime } from "@effect/platform-node";
NodeRuntime.runMain(program.pipe(Effect.provide(AppLayer)));
Error Handling Philosophy
Failures vs defects — the fundamental distinction
| Aspect | Failure (expected) | Defect (unexpected) |
|---|---|---|
| API | Effect.fail(new MyError()) |
Effect.die(new Error()) |
| Type channel | Tracked in E |
Never appears in E (never) |
| Recovery | catchTag, catchAll, retry |
Only at system boundaries |
| Rule of thumb | You intend to handle it at call site | Bug or impossible state |
Always use tagged errors
Plain Error or string failures miss the value of Effect's typed error channel:
class UserNotFound extends Data.TaggedError("UserNotFound")<{
readonly id: string;
}> {}
// Tagged errors are yieldable — no Effect.fail wrapper needed
const program = Effect.gen(function* () {
const user = yield* db.findUser(id);
if (!user) yield* new UserNotFound({ id });
return user;
});
catchAll does NOT catch defects
This is the #1 error handling mistake:
Effect.catchAll(program, handler); // catches E only — NOT defects
Effect.catchAllCause(program, handler); // catches everything (E + defects + interrupts)
Only use catchAllCause / catchAllDefect at system boundaries (top-level error
handlers, HTTP response mappers).
Dependency Injection Architecture
Service → Layer → Provide (once)
1. Define services with Context.Tag → "what do I need?"
2. Implement via Layers → "how is it built?"
3. Provide once at entry point → "wire it all together"
Service methods must have R = never
Dependencies belong in Layer composition, not method signatures:
// WRONG: leaks dependency to callers
findById: (id: string) => Effect.Effect<User, UserNotFound, Database>;
// RIGHT: Database is wired in the Layer
findById: (id: string) => Effect.Effect<User, UserNotFound>;
Layer composition — know the operators
| Operation | When | Behavior |
|---|---|---|
Layer.merge(A, B) |
Independent services | Both build concurrently |
Layer.provide(downstream, upstream) |
A feeds B | upstream builds first |
Layer.fresh(layer) |
Force new instance | Bypasses memoization |
Critical: Layer.merge does NOT sequence construction. If B depends on A, use
Layer.provide, not Layer.merge.
One Effect.provide at the entry point
Scattered provide calls create hidden dependencies and layer duplication:
// WRONG: provide scattered throughout codebase
const getUser = UserRepo.findById(id).pipe(Effect.provide(DbLayer));
// RIGHT: compose and provide once
const main = program.pipe(Effect.provide(AppLayer));
NodeRuntime.runMain(main);
Resource & Scope Rules
Effect.scoped is mandatory for acquireRelease
Forgetting Effect.scoped is the #1 resource management pitfall — resources
accumulate until the program exits:
// WRONG: scope never closes, connection leaks
const result = yield * getDbConnection;
// RIGHT: scope closes when block completes
const result =
yield *
Effect.scoped(
Effect.gen(function* () {
const conn = yield* getDbConnection;
return yield* conn.query("SELECT 1");
}),
);
Release finalizers always run
On success, failure, AND interruption — guaranteed. The finalizer receives the
Exit value for conditional cleanup.
Multiple resources in one scope
Effect.scoped(
Effect.gen(function* () {
const conn = yield* Effect.acquireRelease(openConn(), closeConn);
const file = yield* Effect.acquireRelease(openFile(), closeFile);
// both released when scope closes, in REVERSE acquisition order
}),
);
Concurrency Model
Prefer high-level APIs over raw fork
| API | Use case |
|---|---|
Effect.all([], { concurrency: N }) |
Bounded parallel execution |
Effect.forEach(items, fn, { concurrency: N }) |
Worker pool pattern |
Effect.race(a, b) |
First to complete wins, others interrupted |
Effect.timeout(e, dur) |
Deadline on any effect |
Only reach for Effect.fork / Fiber when high-level APIs are insufficient.
Fork variants — know the lifecycle
| Function | Scope | Cleanup |
|---|---|---|
Effect.fork |
Parent's scope | Auto-interrupted with parent |
Effect.forkDaemon |
Global scope | Nothing cleans it up — you must |
Effect.forkScoped |
Nearest Scope | Tied to resource lifecycle |
Gotcha: forkDaemon leaks fibers if you forget to interrupt them.
Common Pitfalls
-
Floating effects — creating an Effect without yielding or running it is a silent bug.
Effect.log("msg")inside a generator does nothing unlessyield*-ed. -
catchAllwon't catch defects — usecatchAllCauseat system boundaries for full failure visibility. -
Missing
Effect.scoped—acquireReleasewithout a scope boundary leaks resources until program exit. -
Scattered
Effect.provide— compose all layers and provide once at the entry point. -
Point-free on overloaded functions —
Effect.map(myOverloadedFn)silently erases generics. Use explicit lambdas:Effect.map((x) => myOverloadedFn(x)). -
Effect.asyncresume called multiple times — resume must be called exactly once. Multiple calls cause undefined behavior. -
orDiesilences errors — converts typed failures to untyped defects. Handle errors properly instead. -
Layer.mergefor dependent services — merge doesn't sequence construction. UseLayer.providewhen one layer needs another's output. -
Fiber.joinvsFiber.await—joincan cause premature finalizer execution in edge cases. Preferawaitwhen resource safety matters. -
runCollecton infinite streams — never call without a priortake. It will never terminate and consume unbounded memory. -
Using
it.effectfor scoped tests — effects requiringScopemust useit.scoped, notit.effect, or you get a type error.
Reference Files
Read the relevant reference file when working with a specific concern:
| File | When to read |
|---|---|
references/error-handling.md |
Tagged errors, Cause, defect recovery, error mapping patterns |
references/dependency-injection.md |
Services, Layers, composition, memoization, provide patterns |
references/concurrency.md |
Fibers, fork variants, Deferred, Semaphore, structured concurrency |
references/resource-management.md |
Scope, acquireRelease, Layer resources, fork + scope interaction |
references/schema.md |
Schema definition, transforms, branded types, recursive schemas |
references/stream.md |
Stream operators, chunking, backpressure, resourceful streams |
references/testing.md |
@effect/vitest, TestClock, Layer mocking, Config mocking |
references/platform.md |
HTTP client, FileSystem, Command, runtime, framework integration |
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
deps-dev
Look up the latest stable version of any open-source package across npm, PyPI, Go, Cargo, Maven, and NuGet. Use when the user asks 'what's the latest version of X', 'what version should I use', 'is X deprecated', 'how outdated is my package.json/requirements.txt/Cargo.toml', or needs version numbers for adding or updating dependencies. Also covers pinning versions, checking if packages are maintained, or comparing installed vs latest versions. Do NOT use for private/internal packages or for looking up documentation (use context7).
github-codebase-search
Semantic search for public GitHub repos without cloning. Use when the user wants to understand how an external library or framework works internally, investigate upstream bugs, trace code paths in a repo they haven't cloned, or search GitHub source code by intent. Do NOT use for local codebase questions (use codebase-search), documentation lookup (use context7), or private repos.
council-review
Multi-model AI code review — runs Codex, Claude, and Simplify reviews in parallel, then synthesizes a unified report. Use when the user asks to review code changes, audit a diff, check code quality, review a PR, review commits, or review uncommitted changes. Also covers 'code review', 'review my changes', 'check this before I merge', or wanting multiple perspectives on code. Do NOT use for documentation/markdown review or trivial single-line changes.
react-native-advanced
React Native and Expo patterns for navigation, data fetching lifecycle, infinite scroll lists, form handling, state persistence, authentication routing, gesture-driven animations, bottom sheets, push notifications, and OTA updates. Use when building Expo/React Native apps that need screen-level data prefetching, auth guards with protected routes, infinite scroll feeds, native form input handling, offline-capable state persistence, platform-specific setup (focus/online managers), fluid animations and gesture interactions, modal bottom sheets, push notification flows, or over-the-air update strategies. Do not use for React web apps.
react-web-advanced
Web-specific React patterns for type-safe file-based routing, route-level data loading, server-side rendering, search param validation, code splitting, and list virtualization. Use when building React web apps with route loaders, SSR streaming, validated search params, lazy route splitting, or virtualizing large DOM lists. Do not use for React Native apps — use react-native-advanced instead.
context7
Fetch up-to-date documentation for any open-source library or framework. Use when the user asks to look up docs, check an API, find code examples, or verify how a feature works — especially with a specific library name, version migration, or phrases like 'what's the current way to...' or 'the API might have changed'. Also covers setup and configuration docs. Do NOT use for general programming concepts, internal project code, or version lookups (use deps-dev).
Didn't find tool you were looking for?