effect-ts（效果 ts）

Effect-TS

Overview

Provide workflows, patterns, and best practices for building Effect-based TypeScript programs, with focused references for errors, dependencies, resources, runtime execution, generators, schema, testing, platform modules, sink/stream processing, and Effect AI.

Primary docs and API sources:

• https://effect.website/llms.txt (LLM-oriented topic index)
• https://effect.website/docs
• https://effect.website/docs/platform
• https://effect.website/docs/ai/introduction
• https://effect.website/docs/additional-resources/api-reference
• https://tim-smart.github.io/effect-io-ai/ (concise API list)

For internal lookups, the effect-docs MCP can be used to search and fetch API references: https://github.com/tim-smart/effect-mcp.

Progressive Disclosure

1. Start with references/docs-index.md for topic routing.
2. Read one primary guide for the task.
3. Read one adjacent guide only if needed (for cross-cutting concerns like testing, runtime, or observability).
4. Avoid loading unrelated references.

Quick Triage

• If it needs core Effect data types or combinators, open references/core-usage.md.
• If it needs broader data-type choices (DateTime, BigDecimal, HashSet, Redacted), open references/data-types-advanced.md.
• If it needs equality/order/hash semantics, open references/behavior-traits.md.
• If the task is about error modeling or typed failures, open references/error-management.md.
• If it needs error tooling (sandboxing, Cause, error-channel transforms), open references/error-tooling.md.
• If it involves services/dependencies, open references/dependency-management.md.
• If it involves layer construction or test wiring, open references/layer-patterns.md.
• If it involves resource lifecycles, open references/resource-management.md.
• If it involves running effects or runtime choice, open references/runtime-execution.md.
• If it involves fibers or concurrency primitives, open references/concurrency.md.
• If it involves interruption, supervision, or fiber refs, open references/concurrency-advanced.md.
• If it involves schedules or repetition, open references/scheduling.md.
• If it involves retries/backoff or schedule composition, open references/scheduling-retry.md.
• If it involves streams, queues, pubsub, or STM, open references/streams-queues-stm.md.
• If it involves HTTP clients or external APIs, open references/http-client.md.
• If it involves HTTP servers or API definitions, open references/http-server.md.
• If it involves request batching or data loaders, open references/request-resolver.md.
• If it involves caching or memoization, open references/caching.md.
• If it involves configuration or config providers, open references/configuration.md.
• If it involves advanced config or redaction, open references/configuration-advanced.md.
• If it involves logs, metrics, or tracing, open references/observability.md.
• If it needs concrete logger/metrics/tracing setups or exporters, open references/observability-examples.md.
• If it needs wiring of log/metric/trace layers, open references/observability-wiring.md.
• If it needs sequential/branching readability, open references/generators.md.
• If it needs runtime validation/decoding, open references/schema.md.
• If it involves stream consumption patterns with reducers, open references/sink.md.
• If it needs deterministic time in tests, open references/testing.md.
• If it needs broader testing services, open references/testing-stack.md.
• If it involves command/file/path/terminal/key-value modules, open references/platform-primitives.md.
• If it involves LLM workflows, planning, or tool use via Effect AI, open references/ai.md.
• If it involves bundle-size constrained runtimes, open references/micro.md.
• If it involves migrating from Promise/async, open references/migration-async.md.
• If it needs versioning or signature changes, open references/versioning.md.
• If it hits common pitfalls or runtime errors, open references/troubleshooting.md.
• If it needs result inspection or debugging, open references/exit-cause.md.
• If it needs a docs-to-guide map from llms.txt, open references/docs-index.md.

Core Workflow

1. Clarify boundaries and IO; keep core logic as Effect values.
2. Choose style: use pipelines for simple composition; use Effect.gen for sequential logic.
3. Model errors explicitly: type expected errors; treat defects as unexpected failures.
4. Model dependencies with services, tags, and layers; keep interfaces clean of construction concerns.
5. Manage resource lifecycles with Scope when opening/closing resources.
6. Provide the environment via layers and run effects only at the program edge.
7. For platform/infra code, keep side effects in dedicated adapters and expose typed services to domain code.
8. For agent tasks, include a concise rationale for error, concurrency, and runtime choices.

Output Standards

• Show imports and minimal runnable examples.
• Prefer barrel imports (from "effect" / from "@effect/platform") over deep module paths.
• When function-style composition is clearer, use pipe from effect.
• Keep dependency graphs explicit (services, layers, context tags).
• Include error channel types and call out expected vs defect errors.
• Avoid running effects inside libraries; show runtime usage in entrypoints or tests.
• Prefer examples that compile under current Effect major versions.

Agent Quality Checklist

• State the target shape as Effect<A, E, R> when it helps design decisions.
• Explicitly separate expected errors (E) from defects.
• Identify where layers are provided and where run* is called.
• For concurrent code, state bounded/unbounded behavior and shutdown strategy.
• For boundary decoding, show Schema usage and where failures are handled.
• For tests, note which test services are provided (TestContext, TestClock, live overrides).
• For docs uncertainty, consult references/docs-index.md and source docs before finalizing APIs.

References

• Start at references/docs-index.md to choose relevant guides quickly.
• Use references/core-usage.md, references/error-management.md, and references/dependency-management.md as defaults for most code tasks.
• Use platform/runtime references (references/platform-primitives.md, references/runtime-execution.md) for boundary and integration work.
• Use testing/diagnostic references (references/testing.md, references/testing-stack.md, references/troubleshooting.md) for verification and debugging.