Token导航 LogoToken导航

Agent Skill · 研究检索

requirements-engineering(需求工程)

requirements-engineering 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。

requirements-engineering 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,权限为需要联网;许可证为 MIT,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 25,945。

总安装 3,669周安装 147GitHub Stars 636下载 1,188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
MIT

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:requirements-engineering(需求工程)
来源仓库:https://github.com/jasonkneen/kiro
仓库路径:skills/requirements-engineering
安装命令:
npx skills add https://github.com/jasonkneen/kiro --skill requirements-engineering
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/jasonkneen/kiro --skill requirements-engineering

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。

用于查找、检索和筛选相关信息。适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词或任务场景快速定位候选结果。通过 npx skills add 命令从指定仓库安装,需确认权限与维护状态。使用前建议核实是否会触发联网、命令执行或文件读写操作。

requirements-engineering 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/requirements-engineering

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/jasonkneen/kiro --skill requirements-engineeri…;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

    它需要需要联网,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

requirements-engineering 怎么安装?

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

requirements-engineering 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。

requirements-engineering 需要什么权限?

权限级别是需要联网。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。

requirements-engineering 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

requirements-engineering 用的人多吗?

公开数据是累计安装 3,669 次,近一周安装 147 次,下载 1,188 次,GitHub 636 星。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。

requirements-engineering 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。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 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

  • 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。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 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
requirements-engineering

需求工程 · 本页

3,669636需要联网MIT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

Requirements Engineering

Master the art of capturing what needs to be built before diving into how to build it. This skill teaches the EARS (Easy Approach to Requirements Syntax) format for creating clear, testable requirements.

When to Use This Skill

Use requirements engineering when:

  • Starting any new feature or project
  • Clarifying ambiguous stakeholder requests
  • Creating acceptance criteria for user stories
  • Documenting system behavior for testing
  • Ensuring all team members share understanding

The EARS Format

EARS provides consistent patterns for writing requirements that are specific, testable, and unambiguous.

Basic Patterns

Event-Response (Most Common):

WHEN [triggering event] THEN [system] SHALL [required response]

Conditional Behavior:

IF [precondition is met] THEN [system] SHALL [required response]

Complex Conditions:

WHEN [event] AND [additional condition] THEN [system] SHALL [response]

Optional Conditions:

WHEN [event] OR [alternative event] THEN [system] SHALL [response]

Advanced Patterns

State-Based:

WHEN [system is in specific state] THEN [system] SHALL [behavior]

Performance:

WHEN [user action] THEN [system] SHALL [respond within X seconds/milliseconds]

Security:

IF [authentication condition] THEN [system] SHALL [security response]

Step-by-Step Process

Step 1: Capture User Stories

Format: As a [role], I want [feature], so that [benefit]

Focus on:

  • Who is the user? (role)
  • What do they want to accomplish? (feature)
  • Why does it matter? (benefit/value)

Example:

As a returning customer, I want to save my payment methods, so that I can checkout faster in the future.

Step 2: Generate Acceptance Criteria

For each user story, define specific acceptance criteria using EARS:

Example for payment methods:

**User Story:** As a returning customer, I want to save my payment methods, so that I can checkout faster.

**Acceptance Criteria:**
1. WHEN user adds a valid credit card THEN system SHALL securely store card details
2. WHEN user adds a card with invalid number THEN system SHALL display validation error
3. WHEN user has saved cards THEN system SHALL display list during checkout
4. WHEN user selects saved card THEN system SHALL pre-fill payment form
5. WHEN user deletes saved card THEN system SHALL remove card from list
6. IF user is not authenticated THEN system SHALL redirect to login before saving card
7. WHEN user adds card THEN system SHALL mask all but last 4 digits in display

Step 3: Identify Edge Cases

For each requirement, ask:

  • What if the input is empty/null?
  • What if the input is at boundary values?
  • What if the operation fails?
  • What if the user is not authorized?
  • What if there are concurrent operations?

Edge case patterns:

**Error Handling:**
- WHEN [operation fails] THEN system SHALL [display error / retry / log]

**Boundary Conditions:**
- WHEN [value equals minimum/maximum] THEN system SHALL [specific behavior]

**Concurrent Access:**
- WHEN [multiple users access same resource] THEN system SHALL [conflict resolution]

**Empty States:**
- WHEN [collection is empty] THEN system SHALL [display empty state message]

Step 4: Validate Requirements

Use this checklist:

Completeness:

  • All user roles identified and addressed
  • Normal flow scenarios covered
  • Edge cases documented
  • Error cases handled
  • Business rules captured

Clarity:

  • Each requirement uses precise language
  • No ambiguous terms (fast, easy, user-friendly)
  • Technical jargon avoided or defined
  • Expected behaviors are specific

Consistency:

  • EARS format used throughout
  • Terminology consistent across requirements
  • No contradictory requirements
  • Similar scenarios handled similarly

Testability:

  • Each requirement can be verified
  • Success criteria are observable
  • Inputs and expected outputs specified
  • Performance requirements are measurable

Common Mistakes to Avoid

Mistake 1: Vague Requirements

Bad: "System should be fast" Good: "WHEN user submits search THEN system SHALL return results within 2 seconds"

Mistake 2: Implementation Details

Bad: "System shall use Redis for caching" Good: "WHEN user requests frequently accessed data THEN system SHALL return cached results"

Mistake 3: Missing Error Cases

Bad: Only documenting happy path Good: Include WHEN/IF statements for all error conditions

Mistake 4: Untestable Requirements

Bad: "System should be user-friendly" Good: "WHEN new user completes onboarding THEN system SHALL require no more than 3 clicks to reach main dashboard"

Mistake 5: Conflicting Requirements

Bad: Requirements that contradict each other Good: Review all requirements together, resolve conflicts explicitly

Examples

Example 1: File Upload Feature

**User Story:** As a user, I want to upload files, so that I can share documents with my team.

**Acceptance Criteria:**
1. WHEN user selects file under 10MB THEN system SHALL accept file for upload
2. WHEN user selects file over 10MB THEN system SHALL display "file too large (max 10MB)" error
3. WHEN user selects unsupported file type THEN system SHALL display "unsupported format" error with list of allowed types
4. WHEN upload is in progress THEN system SHALL display progress indicator with percentage
5. WHEN upload completes successfully THEN system SHALL display success message with file link
6. WHEN upload fails due to network error THEN system SHALL display retry option
7. IF user is not authenticated THEN system SHALL redirect to login before upload
8. WHEN user uploads file with same name as existing file THEN system SHALL prompt for rename or replace

**Supported File Types:** PDF, DOC, DOCX, XLS, XLSX, PNG, JPG, GIF
**Maximum File Size:** 10MB
**Maximum Files Per Upload:** 5

Example 2: Search Feature

**User Story:** As a customer, I want to search products, so that I can find items quickly.

**Acceptance Criteria:**
1. WHEN user enters search term THEN system SHALL display matching products
2. WHEN search returns results THEN system SHALL show result count
3. WHEN search returns no results THEN system SHALL display "no products found" with suggestions
4. WHEN user searches with special characters THEN system SHALL sanitize input and search
5. WHEN user submits empty search THEN system SHALL display validation message
6. WHEN results exceed 20 items THEN system SHALL paginate with 20 items per page
7. WHEN user searches THEN system SHALL return results within 2 seconds
8. WHEN user types in search box THEN system SHALL show autocomplete suggestions after 3 characters

**Search Fields:** Product name, description, category, SKU
**Minimum Search Length:** 2 characters

Requirements Document Template

# Requirements Document: [Feature Name]

## Overview
[Brief description of the feature and its purpose]

## User Roles
- [Role 1]: [Description of this user type]
- [Role 2]: [Description of this user type]

## Requirements

### Requirement 1: [Name]
**User Story:** As a [role], I want [feature], so that [benefit]

**Acceptance Criteria:**
1. WHEN [event] THEN system SHALL [response]
2. IF [condition] THEN system SHALL [response]
3. WHEN [event] AND [condition] THEN system SHALL [response]

**Edge Cases:**
- [Edge case 1 and how it's handled]
- [Edge case 2 and how it's handled]

### Requirement 2: [Name]
[Continue pattern...]

## Non-Functional Requirements
- **Performance:** [Specific metrics]
- **Security:** [Security requirements]
- **Accessibility:** [Accessibility standards]

## Out of Scope
- [Items explicitly not included in this feature]

## Open Questions
- [Questions that need stakeholder input]

Next Steps

After completing requirements:

  1. Review with stakeholders for accuracy
  2. Get explicit approval before proceeding
  3. Move to Design Phase to create technical architecture
  4. Use requirements as foundation for acceptance testing

平台分布

Claude Code

29.27%
按下载量换算348

Gemini CLI

24.62%
按下载量换算292

OpenCode

17.22%
按下载量换算205

Cursor

12.89%
按下载量换算153

Antigravity

7.71%
按下载量换算92

Codex

3.4%
按下载量换算40

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

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。

继续浏览同类 Skills