Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问许可证需确认审计提醒

scope-task范围任务

Agent Skill

scope-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。

总安装

306

周安装

13

GitHub Stars

公开资料未说明

下载量

107
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:scope-task(范围任务)
来源仓库:https://github.com/b-j-roberts/vibecoding-workshop
仓库路径:skills/scope-task
安装命令:
npx skills add https://github.com/b-j-roberts/vibecoding-workshop --skill scope-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/b-j-roberts/vibecoding-workshop --skill scope-task

简介

用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。

  • 支持基于关键词、任务场景或来源线索进行信息检索与筛选。
  • 可通过 npx skills add 命令从指定 GitHub 仓库安装使用。
  • 安装前需确认权限范围、维护状态及是否触发联网或文件操作。
  • scope-task 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Scope Task

Overview

This skill researches and scopes a single new task, then inserts it into an existing ROADMAP.md at the specified position. It follows the same discovery depth as project-init — web research, self-reflection, and user interview — but focused on one task rather than an entire project. The output is a properly formatted action item inserted into the roadmap, ready for do-task to implement.

Workflow

Step 1: Parse Arguments

Extract two pieces from the user's input:

  • Task ID — numeric identifier in <phase>.<item> format (e.g., 2.5, 3.3)
  • Task Title — the descriptive title following the ID (e.g., "Integrate Research into SPEC.md")

If either is missing or malformed, ask the user to provide the full invocation: /scope-task <phase.item> <Task Title>.

Step 2: Load Project Context

Read the following files from the current working directory (check docs/ subdirectory first, then project root):

  1. ROADMAP.md — to understand existing phases, task numbering, and what has already been scoped
  2. SPEC.md — to understand the project's architecture, tech stack, data models, and key decisions

If ROADMAP.md does not exist, inform the user and stop — this skill requires an existing roadmap to insert into.

Validate the task ID:

  • The phase number must correspond to an existing phase in the roadmap (Phases 1-4 for actionable tasks; Phase 5 for future directions)
  • The item number must not already exist — scope-task always creates new insertions
  • If the ID conflicts with an existing task, list nearby IDs and ask the user to pick an available slot

Step 3: Research

Conduct targeted research using WebSearch to investigate:

  • Best practices — established patterns and approaches for this type of task given the project's tech stack
  • Similar implementations — how other projects have solved this problem
  • Libraries and tools — relevant packages, APIs, or utilities that could aid implementation
  • Trade-offs — alternative approaches with their pros and cons
  • Pitfalls — common mistakes and gotchas specific to this task

Keep research focused on the single task — not the broader project. Use SPEC.md context to make searches specific to the project's stack and architecture.

Step 4: Self-Reflection

Before interviewing the user, pause and reason through:

  • How research findings connect to the project's existing architecture (from SPEC.md)
  • What the task's requirements should be, based on research
  • Gaps in understanding that only the user can fill
  • Potential impacts on other roadmap tasks
  • Scope boundaries — what this task should and should not include
  • Questions where the answer is NOT obvious from SPEC.md, ROADMAP.md, or research

Step 5: Interview

Use AskUserQuestion to clarify unresolved items from Steps 2-4. Topics may include:

  • Technical implementation choices surfaced by research
  • Scope boundaries — what exactly should be in vs. out
  • Integration points with existing or planned roadmap tasks
  • Acceptance criteria the user has in mind
  • Priority or ordering relative to neighboring tasks

Rules for the interview:

  • Do NOT ask questions whose answers are obvious from SPEC.md, ROADMAP.md, or research
  • Do NOT re-ask information already present in the project documentation
  • Each round should surface genuinely new information
  • Keep it focused — this is one task, not a full project discovery
  • 1-2 rounds of questions is typical; go deeper only if the task is genuinely complex

Step 6: Write the Action Item

Compose the new task following the canonical roadmap format.

For Phases 1-4, use exactly 3 subsections:

### <phase>.<item> <Task Title>

**Description**: What this task is and why it matters. Connect it to the project's goals and architecture.

**Requirements**:
- [ ] Concrete deliverable 1
- [ ] Concrete deliverable 2
- [ ] ...

**Implementation Notes**: Technical guidance, references to specific libraries or APIs discovered during research, trade-offs considered, gotchas to watch for. Reference relevant parts of SPEC.md where helpful.

For Phase 5 (Future), use the alternative format:

### <phase>.<item> <Task Title>

**Description**: What this direction is about.

**Features**:
- Potential feature or capability 1
- Potential feature or capability 2

**Rationale**: Why this direction is worth pursuing.

Quality guidelines for the action item:

  • Requirements should be concrete, testable, and individually checkable
  • Implementation Notes should contain genuinely useful technical guidance drawn from research — not generic advice
  • Description should clearly explain *why* this task matters in the context of the project
  • Target 4-10 requirement items depending on task complexity
  • Reference specific libraries, APIs, or patterns by name when applicable

Step 7: Insert into ROADMAP.md

Insert the new action item into ROADMAP.md at the correct position:

  • Place it after the task with the preceding item number (e.g., 2.5 goes after 2.4)
  • If inserting at the end of a phase (e.g., 2.5 when the last task is 2.4), place it before the next phase heading
  • If inserting between existing tasks (e.g., 2.5 between 2.4 and 2.6), place it in numerical order
  • Do NOT renumber existing tasks — the new ID fills a gap or extends the sequence
  • Preserve all existing content and formatting in the roadmap

Use the Edit tool to insert the new task. Verify the insertion by reading back the modified section of ROADMAP.md.

Step 8: Summary

After insertion, present a brief summary to the user:

  • The task ID and title
  • A one-line description of what was scoped
  • The number of requirements defined
  • Any notable trade-offs or decisions made during scoping

The task is now ready for /do-task <id>.

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

37.67%
按下载量换算40

Claude

29.99%
按下载量换算32

Cursor

19.63%
按下载量换算21

Gemini CLI

8.21%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。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