Token导航 LogoToken导航

Agent Skill · 研究检索

prd-generation(珠三角一代)

prd-generation 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。

prd-generation 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,权限为只读;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 48,042。

总安装 706周安装 30GitHub Stars 1下载 247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:prd-generation(珠三角一代)
来源仓库:https://github.com/pixel-process-ug/superkit-agents
仓库路径:skills/prd-generation
安装命令:
npx skills add https://github.com/pixel-process-ug/superkit-agents --skill prd-generation
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/pixel-process-ug/superkit-agents --skill prd-generation

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。

prd-generation 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。适用于产品需求文档相关的信息搜集与筛选,可结合来源仓库和原始 README 核验具体用法。

通过 npx skills add 命令从指定 GitHub 仓库安装,需确认权限范围和维护状态。安装前建议确认是否会触发联网、命令执行或文件读写等操作边界。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。

来源仓库内路径:skills/prd-generation

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/pixel-process-ug/superkit-agents --skill prd-g…;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

prd-generation 怎么安装?

在终端执行 npx skills add https://github.com/pixel-process-ug/superkit-agents --skill prd-generation。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

prd-generation 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。

prd-generation 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

prd-generation 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。

prd-generation 用的人多吗?

公开数据是累计安装 706 次,近一周安装 30 次,下载 247 次,GitHub 1 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。

prd-generation 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。
  • 标注为只读权限,试用阶段不容易误改本地文件。

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。
  • 只想先试试效果、不希望它改动本地文件的人。

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
prd-generation

珠三角一代 · 本页

7061只读未明确通过

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

PRD Generation

Overview

Transform high-level ideas into structured Product Requirements Documents through guided discovery. This skill walks through problem/solution/constraint discovery, generates a professional PRD with measurable goals and user stories, and ensures stakeholder approval before saving.

Announce at start: "I'm using the prd-generation skill to create a Product Requirements Document."

Phase 1: Discovery

Ask these questions ONE AT A TIME (prefer multiple choice where possible).

Do NOT skip discovery. Even if the user provides a detailed brief, confirm understanding by asking at least 3 clarifying questions.

STOP after discovery — present a summary of collected answers and get confirmation before drafting.

Problem Space Questions

#QuestionWhy It Matters
1What problem does this solve?Anchors the entire PRD
2Who are the target users? (personas, roles)Shapes user stories
3How are users currently solving this?Identifies competitive landscape
4What is the impact of NOT solving this?Justifies priority

Solution Space Questions

#QuestionWhy It Matters
5What does success look like? (specific metrics)Defines success metrics
6What are must-have vs nice-to-have features?Sets priority tiers
7What are the explicit non-goals?Prevents scope creep
8Are there existing solutions to learn from?Informs design decisions

Constraint Questions

#QuestionWhy It Matters
9What is the timeline? Any hard deadlines?Scopes release plan
10What technical constraints exist?Narrows solution space
11What resources are available?Sets realistic expectations
12Are there compliance or regulatory requirements?Identifies non-functional reqs

Phase 2: Draft PRD

Generate the PRD using the template below. Dispatch the prd-writer agent with collected answers for heavy generation.

STOP after drafting — do NOT present as final until Phase 3 review is complete.

PRD Template

# [Product/Feature Name] — Product Requirements Document

## 1. Overview
One paragraph summarizing what this is and why it matters.

## 2. Problem Statement
- Current situation
- Pain points
- Impact of not solving

## 3. Goals & Non-Goals
### Goals
- [ ] Goal 1 (measurable)
- [ ] Goal 2 (measurable)

### Non-Goals
- Explicitly NOT doing X
- Explicitly NOT doing Y

## 4. User Stories
As a [persona], I want to [action], so that [benefit].

## 5. Functional Requirements
### FR-1: [Requirement Name]
- Description
- Acceptance criteria
- Priority (P0/P1/P2)

## 6. Non-Functional Requirements
- Performance: [specific targets]
- Security: [requirements]
- Accessibility: [standards]
- Scalability: [expectations]

## 7. Technical Constraints
- Platform/stack requirements
- Integration dependencies
- Data requirements

## 8. Success Metrics
| Metric | Current | Target | How to Measure |
|--------|---------|--------|----------------|

## 9. Timeline & Milestones
| Phase | Description | Target Date |
|-------|-------------|-------------|

## 10. Open Questions
- [ ] Question 1
- [ ] Question 2

## 11. Appendix
References, mockups, related documents

Priority Classification

PriorityMeaningRule
P0Must-have for launchWithout this, the product does not ship
P1Important, ship soon after launchSignificant value but not blocking
P2Nice-to-haveEnhances experience, can wait

Phase 3: Review

Present the PRD section by section:

  1. After each section, ask: "Does this capture your intent? Any changes?"
  2. Revise based on feedback before moving to next section
  3. Pay special attention to these high-signal sections:

- Goals & Non-Goals (scope alignment) - User Stories (persona accuracy) - Success Metrics (measurability) - Functional Requirements (acceptance criteria completeness)

STOP after review — get explicit "approved" confirmation before saving.

Phase 4: Save and Transition

After explicit approval:

  1. Save to docs/prds/YYYY-MM-DD-<feature>.md
  2. Commit the PRD with message: docs(prd): add PRD for <feature>
  3. If implementation follows, invoke the brainstorming skill
  4. If specs are needed, invoke the spec-writing skill

Transition Decision Table

User IntentNext SkillRationale
"Let's build this"brainstorming → planningExplore approaches then plan
"Write the specs"spec-writingBreak PRD into JTBD specs
"Just save it"NonePRD is the deliverable
"Get estimates"task-decompositionBreak into estimable tasks

Anti-Patterns / Common Mistakes

MistakeWhy It Is WrongWhat To Do Instead
Skipping discovery and jumping to draftProduces assumptions-based PRDAlways complete Phase 1 first
Goals without metricsCannot measure successEvery goal needs a number
Missing non-goalsScope creep guaranteedExplicitly list what is out of scope
User stories without acceptance criteriaUntestable requirementsAdd Given/When/Then to each story
Generic success metrics ("improve UX")UnmeasurableUse specific numbers: "reduce load time to <2s"
Presenting entire PRD at once for reviewUser overwhelmed, gives superficial approvalPresent section by section
Copying competitor features verbatimMisses actual user needsFocus on user problems, not solutions

Anti-Rationalization Guards

  • Do NOT skip discovery because "the user already described it well enough"
  • Do NOT leave placeholder text in any section — fill every section or mark as "TBD: [reason]"
  • Do NOT proceed to save without explicit user approval of each section
  • Do NOT invent success metrics — they must come from the user

Integration Points

SkillRelationship
brainstormingUpstream: explores ideas before PRD; downstream: explores implementation after PRD
spec-writingDownstream: PRD provides high-level requirements; specs detail them with JTBD
planningDownstream: plan references PRD requirements for task breakdown
task-decompositionDownstream: breaks PRD into estimable work items
tech-docs-generatorParallel: PRD informs what documentation is needed
acceptance-testingDownstream: acceptance criteria from PRD feed test definitions

Verification Gate

Before claiming the PRD is complete:

  1. VERIFY all 11 sections are filled (not placeholder text)
  2. VERIFY every goal has a measurable metric
  3. VERIFY non-goals are explicit and meaningful
  4. VERIFY user stories have acceptance criteria
  5. VERIFY user has approved each section individually
  6. VERIFY the file is saved and committed

Concrete Example: Discovery Summary

Problem: Users cannot find relevant search results in the dashboard.
Users: Data analysts (primary), team leads (secondary).
Current workaround: Export to Excel and use Ctrl+F.
Impact of not solving: 30min/day wasted per analyst (team of 12).
Success metric: Reduce average search time from 5min to <30s.
Must-have: Full-text search across all dashboard widgets.
Non-goal: Advanced boolean query syntax (P2, not launch).
Timeline: 6 weeks to MVP.
Constraint: Must work with existing Elasticsearch cluster.

This summary is presented to the user for confirmation before Phase 2 begins.

Skill Type

Flexible — Adapt discovery depth and PRD structure to project context while preserving the discovery-before-drafting principle and section-by-section review process.

平台分布

Codex

35.45%
按下载量换算88

Claude

34.53%
按下载量换算85

Cursor

17.67%
按下载量换算44

Gemini CLI

8.85%
按下载量换算22

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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