Token导航 LogoToken导航

Agent Skill · 研究检索

requirements-elicitation(需求启发)

requirements-elicitation 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-elicitation 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 66,724。

总安装 259周安装 11GitHub Stars 265下载 91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-elicitation(需求启发)
来源仓库:https://github.com/rsmdt/the-startup
仓库路径:skills/requirements-elicitation
安装命令:
npx skills add https://github.com/rsmdt/the-startup --skill requirements-elicitation
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/rsmdt/the-startup --skill requirements-elicitation

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-elicitation 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-elicitation

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/rsmdt/the-startup --skill requirements-elicita…;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

requirements-elicitation 怎么安装?

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

requirements-elicitation 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-elicitation 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。

requirements-elicitation 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-elicitation 用的人多吗?

公开数据是累计安装 259 次,近一周安装 11 次,下载 91 次,GitHub 265 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-elicitation 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-elicitation

需求启发 · 本页

259265操作浏览器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 Elicitation Methodology

Systematic techniques for transforming vague ideas into clear, testable specifications that align stakeholders and guide implementation.

When to Activate

  • Gathering requirements for new features
  • Clarifying vague or ambiguous requests
  • Resolving conflicting stakeholder needs
  • Documenting formal specifications
  • Creating user stories with acceptance criteria
  • Validating requirements before implementation

Elicitation Techniques

The 5 Whys

Drill past surface requests to discover root needs:

Surface Request: "We need a dashboard"

Why 1: Why do you need a dashboard?
→ "To see our metrics in one place"

Why 2: Why do you need to see metrics in one place?
→ "To identify problems quickly"

Why 3: Why do you need to identify problems quickly?
→ "Because slow response affects customer satisfaction"

Why 4: Why does customer satisfaction matter right now?
→ "We're losing customers and don't know why until it's too late"

Why 5: Why don't you know until it's too late?
→ "We only see issues in monthly reports"

Root Need: Real-time alerting for customer-impacting issues
(Not just a dashboard - the dashboard was a solution, not the need)

Concrete Examples

Transform abstract requirements into specific, testable scenarios:

AbstractConcrete
"The system should be fast""Page loads in under 2 seconds on 3G"
"Users should be able to search""Find orders by customer name, date range, or status"
"It needs to be secure""All PII encrypted at rest, session timeout after 15 min inactive"
"Good error handling""Network failures retry 3x with exponential backoff, then show offline mode"

Boundary Identification

Define what's explicitly in and out of scope:

Feature: User Registration

IN SCOPE:
✓ Email/password registration
✓ Email verification
✓ Password strength requirements
✓ Terms of service acceptance

OUT OF SCOPE:
✗ Social login (Google, Facebook)
✗ Two-factor authentication
✗ Password recovery (separate feature)

DEFERRED:
◐ SSO integration (planned for Q3)
◐ Biometric login (pending security review)

Stakeholder Interviews

Structured conversation to extract requirements:

Interview Structure (45 min):

1. CONTEXT (10 min)
   - What's your role in this project?
   - What does success look like for you?
   - What's driving this initiative?

2. CURRENT STATE (10 min)
   - How do you do this today?
   - What works well?
   - What are the pain points?

3. DESIRED STATE (15 min)
   - What would the ideal solution look like?
   - Walk me through a typical scenario...
   - What would make your job easier?

4. CONSTRAINTS (5 min)
   - What absolutely must be included?
   - What's definitely out of scope?
   - Any timeline or budget constraints?

5. WRAP-UP (5 min)
   - What haven't I asked that I should?
   - Who else should I talk to?
   - Can I follow up if I have questions?

Observation and Shadowing

Watch users perform tasks in their environment:

Observation Protocol:

PREPARE:
- Define what you're observing
- Get permission to observe
- Prepare note-taking template

OBSERVE:
- Note actions, not interpretations
- Record workarounds and pain points
- Note environmental factors
- Time key activities

DEBRIEF:
- "I noticed you did X, can you tell me more?"
- "What would make that easier?"
- "How often does this happen?"

Document:
┌─────────────────────────────────────────────────────────────┐
│ Observation: Order Processing                               │
├─────────────────────────────────────────────────────────────┤
│ Action: Copied customer email from order to support tool    │
│ Time: 15 seconds per order                                  │
│ Frequency: ~50 orders/day                                   │
│ Pain Point: Manual copy-paste, error-prone                  │
│ Opportunity: Direct integration between systems             │
└─────────────────────────────────────────────────────────────┘

Requirement Documentation

User Story Format

Format:
As a [role],
I want [capability],
So that [benefit].

Components:
- Role: Who benefits (be specific)
- Capability: What they can do (action, not solution)
- Benefit: Why it matters (business value)

Example:
As a customer service representative,
I want to see a customer's order history when they call,
So that I can resolve their issues without asking them to repeat information.

Acceptance Criteria (Given-When-Then)

Format:
Given [context/precondition]
When [action/event]
Then [expected outcome]

Example:
Feature: Order Cancellation

Scenario: Cancel order before shipping
Given an order in "confirmed" status
And the order has not been shipped
When the customer requests cancellation
Then the order status changes to "cancelled"
And the customer receives a cancellation confirmation email
And the payment is refunded within 3-5 business days

Scenario: Cannot cancel shipped order
Given an order in "shipped" status
When the customer requests cancellation
Then the cancellation is rejected
And the customer is directed to the returns process

Edge Cases and Exceptions

Document what happens when things go wrong:

Feature: Password Reset

Happy Path:
- User requests reset → Email sent → User clicks link → Sets new password

Edge Cases:
| Scenario | Expected Behavior |
|----------|-------------------|
| Email not found | Show same success message (security) |
| Link expired (>24h) | Show "link expired" with new reset option |
| Link already used | Show "link already used" message |
| Weak password | Show requirements, block submission |
| Same as old password | Show error, require different password |
| User locked out | Still send reset email (unlock flow) |

Non-Functional Requirements

Document quality attributes:

NFR Template:
┌─────────────────────────────────────────────────────────────┐
│ Category: Performance                                        │
├─────────────────────────────────────────────────────────────┤
│ Requirement: Response Time                                   │
│ Measure: 95th percentile page load time                     │
│ Target: < 2 seconds                                          │
│ Context: Desktop browser, 4G connection                     │
│ Priority: Must Have                                          │
└─────────────────────────────────────────────────────────────┘

Common Categories:
- Performance: Speed, throughput, latency
- Scalability: Users, data volume, geographic distribution
- Availability: Uptime, recovery time, disaster recovery
- Security: Authentication, authorization, encryption
- Usability: Accessibility, learnability, efficiency
- Maintainability: Modularity, testability, documentation

Stakeholder Management

Stakeholder Analysis

Identify and categorize stakeholders:

Stakeholder Map:

             High Influence
                   │
    ┌──────────────┼──────────────┐
    │   Manage     │    Partner   │
    │   Closely    │    With      │
    │              │              │
Low ├──────────────┼──────────────┤ High
Interest          │              Interest
    │   Monitor    │    Keep      │
    │   Only       │    Informed  │
    │              │              │
    └──────────────┼──────────────┘
                   │
             Low Influence

Stakeholder Register:
| Name | Role | Interest | Influence | Communication |
|------|------|----------|-----------|---------------|
| VP Sales | Sponsor | High | High | Weekly update |
| Dev Team | Implementer | High | Medium | Daily standup |
| Legal | Advisor | Low | High | As needed |

RACI Matrix

Define roles for each requirement:

R = Responsible (does the work)
A = Accountable (final decision maker)
C = Consulted (provides input)
I = Informed (kept updated)

| Requirement | Product | Dev | Design | Legal |
|-------------|---------|-----|--------|-------|
| User stories | R,A | C | C | I |
| UI mockups | C | I | R,A | I |
| API contracts | C | R,A | I | I |
| Privacy policy | C | I | I | R,A |

Conflict Resolution

When stakeholders disagree:

Resolution Process:

1. UNDERSTAND both positions
   - "Help me understand why X is important to you"
   - Identify underlying needs vs stated positions

2. FIND COMMON GROUND
   - What do both parties agree on?
   - What's the shared goal?

3. EXPLORE OPTIONS
   - Can we do both? (phased approach)
   - Is there a third option that addresses both needs?
   - What's the minimum viable for each?

4. ESCALATE if needed
   - Present options with trade-offs
   - Let decision-maker decide
   - Document the decision and rationale

Example:
Marketing wants: Launch by Q1 with all features
Engineering says: Can't do all features by Q1

Resolution: Launch Q1 with core features (MVP), Phase 2 in Q2
Documented: ADR-2024-03: MVP Scope Decision

Validation Techniques

Requirements Review Checklist

CriterionQuestionPass/Fail
CompleteIs everything needed documented?
ConsistentAre there contradictions?
CorrectDoes it match stakeholder intent?
UnambiguousIs there only one interpretation?
TestableCan we verify it's met?
TraceableCan we link to business goal?
FeasibleCan it be implemented?
PrioritizedIs importance clear?

Prototype Validation

Use prototypes to validate understanding:

Prototype Levels:

Low Fidelity (Paper/Whiteboard):
- Quick to create (minutes)
- Good for: Overall flow, major screens
- Validate: "Is this the right approach?"

Medium Fidelity (Clickable mockups):
- Moderate effort (hours)
- Good for: Detailed interactions, UI layout
- Validate: "Does this workflow make sense?"

High Fidelity (Functional prototype):
- Significant effort (days)
- Good for: Complex interactions, performance
- Validate: "Will this actually work?"

Acceptance Criteria Review

Validate with stakeholders before implementation:

Review Format:

"Here's my understanding of [feature]. Please correct me if I'm wrong."

[Read each scenario aloud]

Questions:
- "Is this what you expected?"
- "What did I miss?"
- "What edge cases should we handle?"
- "Is the priority right?"

Document changes and get sign-off.

Traceability

Traceability Matrix

Link requirements to their sources and verification:

| Req ID | Description | Source | Priority | Status | Test Cases |
|--------|-------------|--------|----------|--------|------------|
| REQ-001 | User login | Stakeholder interview | Must | Approved | TC-001, TC-002 |
| REQ-002 | Order history | User observation | Should | Draft | TC-015 |
| REQ-003 | Export CSV | Sales team request | Could | Approved | TC-020 |

Requirement States

Track requirement lifecycle:

States:
┌─────────┐     ┌──────────┐     ┌──────────┐     ┌────────────┐
│ Draft   │────►│ Reviewed │────►│ Approved │────►│ Implemented│
└─────────┘     └──────────┘     └──────────┘     └────────────┘
                     │                                   │
                     ▼                                   ▼
                ┌──────────┐                      ┌──────────┐
                │ Rejected │                      │ Verified │
                └──────────┘                      └──────────┘

Anti-Patterns

Anti-PatternProblemSolution
Solution First"We need a dashboard"Ask "Why?" to find the real need
Assumed ObviousUndocumented "common sense"Document everything explicitly
Gold PlatingAdding unrequested featuresStick to documented requirements
Moving BaselineRequirements keep changingChange control process
Single StakeholderMissing perspectivesIdentify all stakeholders
Technical JargonUsers don't understandUse domain language

Templates

Feature Request Template

# Feature: [Name]

## Problem Statement
[What problem does this solve?]

## User Stories
- As a [role], I want [what] so that [why]

## Acceptance Criteria
- Given [context] when [action] then [outcome]

## Out of Scope
- [What this feature does NOT include]

## Dependencies
- [Other features or systems required]

## Open Questions
- [Unresolved issues needing discussion]

Requirements Document Template

# [Project Name] Requirements Specification

## 1. Introduction
### 1.1 Purpose
### 1.2 Scope
### 1.3 Definitions

## 2. Overall Description
### 2.1 Product Perspective
### 2.2 User Classes
### 2.3 Constraints

## 3. Functional Requirements
### 3.1 [Feature Area 1]
### 3.2 [Feature Area 2]

## 4. Non-Functional Requirements
### 4.1 Performance
### 4.2 Security
### 4.3 Usability

## 5. Appendices
### A. Stakeholder Register
### B. Traceability Matrix

References

平台分布

Claude Code

29.35%
按下载量换算27

OpenCode

21.21%
按下载量换算19

windsurf

15.56%
按下载量换算14

trae

12.53%
按下载量换算11

Gemini CLI

6.71%
按下载量换算6

Antigravity

3.11%
按下载量换算3

安全审计

Gen Agent Trust Hub

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。

继续浏览同类 Skills