Token导航 LogoToken导航

Agent Skill · 研究检索

task(任务管理)

task 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。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。

task 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为操作浏览器;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 57,063。

总安装 441周安装 18GitHub Stars 公开资料未说明下载 143核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:task(任务管理)
来源仓库:https://github.com/eljun/claude-skills
仓库路径:skills/task
安装命令:
npx skills add https://github.com/eljun/claude-skills --skill task
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/eljun/claude-skills --skill task

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于任务规划与需求澄清。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。适合在 Codex、Claude、Cursor、Gemini CLI 中拆分工作项。自动分配 Task ID 并创建标准化文档模板。

支持 auto-pipeline 自动进入实现阶段。task 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/task

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/eljun/claude-skills --skill task;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要操作浏览器,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

task 怎么安装?

在终端执行 npx skills add https://github.com/eljun/claude-skills --skill task。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

task 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

task 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

task 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

task 用的人多吗?

公开数据是累计安装 441 次,近一周安装 18 次,下载 143 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

task 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
task

任务管理 · 本页

441—操作浏览器未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

/task - Task Planning Agent

Model: opus (complex planning requires advanced reasoning)

Command Flags

FlagShortDescription
--help-hShow available commands and options
--version-vShow workflow skills version
autoEnable automated pipeline (implement → test → document → ship)

Flag Handling

On -h or --help:

/task - Task Planning Agent

Usage:
  /task                    Create a new task (manual mode)
  /task auto               Create task with auto-pipeline enabled
  /task -h, --help         Show this help message
  /task -v, --version      Show version

Options:
  auto    After task approval, automatically chain through:
          implement → test → document → ship

Examples:
  /task                    # Interactive task creation
  /task auto               # Task with full automation

Next: /implement {ID}

On -v or --version: Display:

Workflow Skills v1.4.1
https://github.com/eljun/claude-skills

When to Use

Invoke /task when:

  • Starting a new feature or enhancement
  • Planning a bug fix that requires multiple changes
  • User says "I want to add...", "Let's implement...", "Can we build..."
  • Discussing requirements before coding
  • Creating implementation specs for future work

Invocation Options

CommandModeBehavior
/taskManualYou control each step: implement → test → document → ship
/task autoAutomatedAfter task approval, all steps run autonomously

Auto Mode Workflow

/task auto → User approves task
    ↓
/implement
    ↓
/test (Playwright E2E)
    │
PASS → /document
FAIL → /implement (with test report)
    │
    ▼
/ship → PR + notify

Auto mode notes:

  • Full automation: Runs through implement → test → document → ship
  • Test failures: Auto-retries by sending test report back to implement agent
  • Unexpected errors: Stops and notifies you

Workflow

User Request
     ↓
1. Discuss & clarify requirements
2. Research codebase for context
3. Generate Task ID (next available number)
4. Create docs/task/{ID}-{task-name}.md
5. Add to TASKS.md under "## Planned" with ID
6. User can /clear and start fresh
     ↓
Ready for /implement {ID}

What This Skill Does

1. Requirements Gathering

Ask clarifying questions:

  • What is the expected behavior?
  • Any UI/UX preferences?
  • Priority level?

Version Impact Guidelines

Set the Version Impact field based on the type of change:

TypeDefault ImpactWhen to Use
featureminorNew functionality, backwards compatible
bugfixpatchBug fixes, no new features
enhancementpatchImprovements to existing features
documentationpatchDoc updates only
chorepatchMaintenance, refactoring

Override to major when:

  • Breaking API changes
  • Database schema changes requiring migration
  • Removing deprecated features
  • Changes that require user action

2. Codebase Research

Before creating the task document:

  • Check existing similar implementations
  • Identify files to modify
  • Find reusable components/hooks
  • Note potential pitfalls
Note: Specialized skills (vercel-react-best-practices, supabase-postgres-best-practices) are invoked during /implement, not during task planning. This keeps planning focused on requirements and architecture.

3. Generate Task ID

Before creating the task document, generate the next available Task ID:

  1. Read TASKS.md to find all existing task IDs
  2. Find the highest ID number across all sections (Planned, In Progress, Testing, etc.)
  3. Assign next ID = highest + 1
  4. If no tasks exist, start with ID = 1

ID Format: Simple integers (1, 2, 3,...)

# Example: If TASKS.md has tasks with IDs 1, 2, 5
# Next ID = 6 (highest + 1)

4. Create Task Document

Location: docs/task/{ID}-{task-name}.md

Naming: Zero-padded ID prefix + kebab-case descriptive name:

  • 001-user-dashboard-redesign.md
  • 002-lead-auto-tagging.md
  • 003-booking-calendar-view.md

ID Padding: Use 3 digits (001, 002,... 999) for consistent sorting.

5. Update TASKS.md

Add the task to the "## Planned" section with ID and link to task document.

If TASKS.md doesn't exist, create it first with this structure:

# Tasks

Task tracking for the development workflow.

---

## Planned

Tasks ready for `/implement {ID}`.

| ID | Task | Priority | Task Doc | Created |
|----|------|----------|----------|---------|

---

## In Progress

| ID | Task | Started | Task Doc | Status |
|----|------|---------|----------|--------|

---

## Testing

Tasks being tested via `/test`.

| ID | Task | Task Doc | Test Report | Status |
|----|------|----------|-------------|--------|

---

## Approved

Tested and approved. Ready for `/document` then `/ship`.

| ID | Task | Task Doc | Feature Doc | Test Report | Approved |
|----|------|----------|-------------|-------------|----------|

---

## Ready to Ship

PRs created via `/ship`. **Items stay here until `/release` is run** (even after merge).

| ID | Task | Branch | PR | Merged | Task Doc |
|----|------|--------|----|--------|----------|

---

## Shipped

Released items. Only `/release` moves items here with version number.

| ID | Task | PR | Release | Shipped |
|----|------|-----|---------|---------|

Task Document Template

Create this structure in docs/task/{ID}-{task-name}.md:

# {Task Title}

> **ID:** {number}
> **Status:** PLANNED
> **Priority:** HIGH | MEDIUM | LOW
> **Type:** feature | bugfix | enhancement | documentation | chore
> **Version Impact:** minor | patch | major
> **Created:** {Date}
> **Platform:** Web
> **Automation:** manual | auto

## Overview

{2-3 sentence description of what we're building and why}

## Requirements

### Must Have
- [ ] Requirement 1
- [ ] Requirement 2

### Nice to Have
- [ ] Optional requirement

## Current State

{Description of how things work now, if applicable}

**Current Files:**
| File | Purpose |
|------|---------|
| `path/to/file.tsx` | Description |

## Proposed Solution

{Description of the implementation approach}

### Architecture

{High-level design decisions}

### File Changes

| Action | File | Description |
|--------|------|-------------|
| CREATE | `path/to/new.tsx` | New component for X |
| MODIFY | `path/to/existing.tsx` | Add Y functionality |
| DELETE | `path/to/old.tsx` | No longer needed |

## Implementation Steps

### Step 1: {Title}
{Detailed instructions with code snippets if needed}

### Step 2: {Title}
{Detailed instructions}

## Code Examples

{Include specific code changes when helpful}

// Example of key implementation


## Testing Checklist

- Test case 1
- Test case 2
- Edge case handling

## Dependencies

- Required packages: {list any new deps}
- Required APIs: {list endpoints needed}
- Blocked by: {any dependencies on other tasks}

## Notes for Implementation Agent

{Any important context the /implement agent needs to know}

## Related

- Similar feature: [link to docs]
- Design reference: [link if applicable]

TASKS.md Integration

After creating the task document, add an entry with the ID:

## Planned

| ID | Task | Priority | Task Doc | Created |
|----|------|----------|----------|---------|
| 1 | Dashboard Redesign | HIGH | [001-dashboard-redesign.md](docs/task/001-dashboard-redesign.md) | Jan 25 |
| 2 | Fix Login Bug | MEDIUM | [002-fix-login-bug.md](docs/task/002-fix-login-bug.md) | Jan 26 |

Output Checklist

Before completing /task:

  • Task document created in docs/task/
  • Document has all required sections filled
  • Implementation steps are clear and actionable
  • File paths verified to exist (for modifications)
  • Added to TASKS.md "## Planned" section
  • User understands the task and approves

Handoff to /implement

When planning is complete, inform the user:

Manual Mode

Task created: #{ID} - {Task Title}
Document: docs/task/{ID}-{task-name}.md
Added to TASKS.md under "Planned"

Next Steps:
  /implement {ID}              # e.g., /implement 1
  /implement {ID}-{task-name}  # e.g., /implement 001-auth-jwt

(Optional: /clear first to start fresh session)

Auto Mode

When /task auto was invoked and user approves the task:

  1. Set Automation: auto in the task document
  2. Use Task tool to spawn /implement {ID} with model: opus
  3. The implement skill will chain to subsequent skills automatically
Task approved! Starting automated pipeline...
Task: #{ID} - {Task Title}

Spawning /implement {ID} with opus model...

IMPORTANT: In auto mode, after user approves the task:

  • Do NOT wait for user to invoke /implement
  • Use Task tool to spawn implement agent with model: opus
  • Example: Task({subagent_type: "general-purpose", model: "opus", prompt: "/implement {ID}"})
  • The automation flag in the task doc controls subsequent chaining

Related Skills

SkillWhen to Use
/implementAfter task is approved, to start coding
/testAfter implementation, to verify the feature
/documentAfter tests pass, to create documentation
/shipAfter documentation, to create PR
/releaseAfter multiple items shipped, to create release

Note: Specialized skills (vercel-react-best-practices, supabase-postgres-best-practices) are invoked during /implement, not during /task. Install them separately from their respective repos.

平台分布

Codex

34.76%
按下载量换算50

Claude

33.17%
按下载量换算47

Cursor

19.32%
按下载量换算28

Gemini CLI

10.57%
按下载量换算15

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

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。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills