Token导航 LogoToken导航TokenDH.com
开发执行命令github未标认证来源可访问许可证需确认审计通过

dart-concurrency-isolatesdart 并发隔离

Agent Skill

dart-concurrency-isolates 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

1,536

周安装

64

GitHub Stars

61

下载量

512
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:dart-concurrency-isolates(dart 并发隔离)
来源仓库:https://github.com/dart-lang/skills
仓库路径:skills/dart-concurrency-isolates
安装命令:
npx skills add https://github.com/dart-lang/skills --skill dart-concurrency-isolates
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/dart-lang/skills --skill dart-concurrency-isolates

简介

dart-concurrency-isolates 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息。

  • 适合围绕仓库状态、代码变更或协作事项进行整理。
  • 可结合来源仓库和原始 README 核验具体用法。
  • 安装前建议确认权限范围、维护状态及是否触发联网或文件读写。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

Managing Dart Concurrency and Isolates

Contents

Core Guidelines

  • Isolate Memory: Assume zero shared memory between isolates. Isolates communicate exclusively via message passing.
  • Data Transfer: Avoid passing large mutable objects between isolates. Prefer simple data types or immutable records to minimize serialization overhead.
  • Resource Management: Always ensure isolates and ports are terminated when no longer needed to prevent memory leaks.
  • Platform Limitations: Do not use isolates on the Dart Web platform. Web compiles to JavaScript, which uses Web Workers instead.
  • Related Skills: Refer to dart-async-programming for standard asynchronous operations (Future, Stream, async/await) running on the Main Isolate.

Choosing the Right Isolate Strategy

Apply conditional logic to determine the correct isolate implementation:

  • If executing a simple, one-off background task (e.g., parsing a single large JSON payload, compressing a file): Use Isolate.run().
  • If executing complex, long-running background workers (e.g., continuous data processing, maintaining a persistent database connection): Use Isolate.spawn() and manually manage SendPort and ReceivePort.

Implementing One-Off Tasks

Use Isolate.run() to spawn an isolate, execute a function, capture the result, and automatically terminate the isolate.

  1. Pass a top-level function, static method, or closure to Isolate.run().
  2. Await the result in the Main Isolate.
Future<Map<String, dynamic>> parseLargeJson(String jsonString) async {
  // Spawns isolate, runs decode, returns result, and terminates automatically.
  return await Isolate.run(() => jsonDecode(jsonString) as Map<String, dynamic>);
}

Implementing Long-Running Workers

Manually manage SendPort and ReceivePort to establish two-way communication for long-lived worker isolates.

  1. Initialize with RawReceivePort: Use RawReceivePort in the Main Isolate to separate startup logic from ongoing message handling.
  2. Establish Two-Way Communication: Pass the Main Isolate's SendPort to the Worker Isolate via Isolate.spawn(). The Worker Isolate must create its own ReceivePort and send its SendPort back to the Main Isolate.
  3. Map Requests to Responses: Use a Completer map with unique IDs to track asynchronous requests sent to the Worker Isolate and resolve them when the response is received.
  4. Handle Errors: Catch exceptions in the Worker Isolate and send them back as RemoteError objects.
  5. Teardown: Send a specific shutdown message to the Worker Isolate to close its ReceivePort, and close the Main Isolate's ReceivePort when all active requests are completed.

Workflows

Task Progress: Long-Running Worker Setup

Copy this checklist to track progress when implementing a long-running worker isolate:

  • Create a Worker class to encapsulate isolate management.
  • Implement a static spawn() method using RawReceivePort to capture the initial SendPort from the worker.
  • Call Isolate.spawn(), passing the RawReceivePort.sendPort and the worker entrypoint method.
  • Implement the worker entrypoint method (_startRemoteIsolate).
  • Create a ReceivePort inside the worker and send its SendPort back to the Main Isolate.
  • Set up a listener on the worker's ReceivePort to process incoming commands.
  • Set up a listener on the Main Isolate's ReceivePort to process responses and resolve Completer instances.
  • Implement a close() method to send a shutdown command and close all ports.
  • Run validator -> review errors -> fix (Ensure no memory leaks and all ports close cleanly).

Examples

Robust Long-Running Worker Implementation

Use this pattern for robust, two-way communication with a long-running worker isolate.

import 'dart:async';
import 'dart:convert';
import 'dart:isolate';

class JsonWorker {
  final SendPort _commands;
  final ReceivePort _responses;
  final Map<int, Completer<Object?>> _activeRequests = {};
  int _idCounter = 0;
  bool _closed = false;

  JsonWorker._(this._responses, this._commands) {
    _responses.listen(_handleResponsesFromIsolate);
  }

  static Future<JsonWorker> spawn() async {
    final initPort = RawReceivePort();
    final connection = Completer<(ReceivePort, SendPort)>.sync();

    initPort.handler = (initialMessage) {
      final commandPort = initialMessage as SendPort;
      connection.complete((
        ReceivePort.fromRawReceivePort(initPort),
        commandPort,
      ));
    };

    try {
      await Isolate.spawn(_startRemoteIsolate, initPort.sendPort);
    } catch (e) {
      initPort.close();
      rethrow;
    }

    final (ReceivePort receivePort, SendPort sendPort) = await connection.future;
    return JsonWorker._(receivePort, sendPort);
  }

  Future<Object?> parseJson(String message) async {
    if (_closed) throw StateError('Worker is closed');
    final completer = Completer<Object?>.sync();
    final id = _idCounter++;
    _activeRequests[id] = completer;
    _commands.send((id, message));
    return await completer.future;
  }

  void _handleResponsesFromIsolate(dynamic message) {
    final (int id, Object? response) = message as (int, Object?);
    final completer = _activeRequests.remove(id)!;

    if (response is RemoteError) {
      completer.completeError(response);
    } else {
      completer.complete(response);
    }

    if (_closed && _activeRequests.isEmpty) _responses.close();
  }

  static void _startRemoteIsolate(SendPort sendPort) {
    final receivePort = ReceivePort();
    sendPort.send(receivePort.sendPort);

    receivePort.listen((message) {
      if (message == 'shutdown') {
        receivePort.close();
        return;
      }

      final (int id, String jsonText) = message as (int, String);
      try {
        final jsonData = jsonDecode(jsonText);
        sendPort.send((id, jsonData));
      } catch (e) {
        sendPort.send((id, RemoteError(e.toString(), '')));
      }
    });
  }

  void close() {
    if (!_closed) {
      _closed = true;
      _commands.send('shutdown');
      if (_activeRequests.isEmpty) _responses.close();
    }
  }
}

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

38.59%
按下载量换算198

Claude

28.76%
按下载量换算147

Cursor

19.38%
按下载量换算99

Gemini CLI

9.05%
按下载量换算46

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/dart-lang/skills --skill dart-concurrency-isolates 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills