Token导航 LogoToken导航

Agent Skill · 研究检索

prd-writing(PRD 写作)

用于辅助文档、README、Markdown、说明文和内容稿件的整理与改写。它适合让 Agent 提炼结构、补齐章节、统一术语、检查链接或把零散材料整理成可读文档。使用时应保留项目已有事实、命令和路径,不要把未确认的信息写成确定结论;涉及对外文案时,还需要控制语气,避免过度营销或夸大能力。

prd-writing 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 44,563。

总安装 906周安装 37GitHub Stars 114下载 293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-writing(PRD 写作)
来源仓库:https://github.com/shawnpang/startup-founder-skills
仓库路径:skills/prd-writing
安装命令:
npx skills add https://github.com/shawnpang/startup-founder-skills --skill prd-writing
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/shawnpang/startup-founder-skills --skill prd-writing

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-writing 用于辅助文档、README、Markdown、说明文和内容稿件的整理与改写。它适合让 Agent 提炼结构、补齐章节、统一术语、检查链接或把零散材料整理成可读文档。使用时保留项目已有事实、命令和路径,不要把未确认的信息写成确定结论。涉及对外文案时,还需要控制语气,避免过度营销或夸大能力。

当前底部简介为空,需结合原始 SKILL.md 内容进一步核验功能细节和使用限制。

来源仓库内路径:skills/prd-writing

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/shawnpang/startup-founder-skills --skill prd-w…;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-writing 怎么安装?

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

prd-writing 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-writing 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-writing 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-writing 用的人多吗?

公开数据是累计安装 906 次,近一周安装 37 次,下载 293 次,GitHub 114 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-writing 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-writing

PRD 写作 · 本页

906114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 Writing

When to Use

Activate when a founder or PM needs to turn a product idea, feature request, or strategic initiative into a structured Product Requirements Document. This includes situations where the user says things like "write a PRD," "spec out this feature," "define requirements for X," or "I need to document what we're building."

Context Required

  • From startup-context: company stage, target customer segments, current product state, team size, technical constraints.
  • From the user: the feature or initiative to spec, known user problems it addresses, any prior research or customer feedback, desired timeline, and scope preference (lightweight vs. full PRD).

Workflow

  1. Clarify scope level -- Ask whether this needs a lightweight PRD (early-stage exploration, 2-3 pages) or a full PRD (committed initiative, 5-8 pages). Default to lightweight if the company is pre-product-market-fit.
  2. Gather inputs -- Collect the problem statement, target users, any existing research, success criteria, and known constraints. Identify key contacts and their roles.
  3. Draft the 8-section PRD -- Write each section sequentially using the template below. Use accessible language suitable for a broad audience including engineering, design, and leadership.
  4. Flag assumptions -- Explicitly list key assumptions underlying each section. For each, state what evidence supports it and what would invalidate it.
  5. Review and refine -- Present the draft, invite feedback, and iterate on specific sections. State the PRD version and last-updated date.

Output Format

A structured PRD document with 8 sections:

Section Template

  1. Summary -- 2-3 sentence overview of what is being built and why it matters. Write for a broad audience.
  2. Contacts -- Key stakeholders with their roles and relevant context about their involvement.
  3. Background -- Context on the problem space: why now, what changed, what enables this initiative. Include competitive context on how others handle the same problem.
  4. Objective -- Goals, business and customer benefits, and strategic alignment. Define SMART success metrics tied to OKRs. Use the format: "Enable [user segment] to [action] resulting in [measurable outcome]."
  5. Market Segment(s) -- Define target users by problems and needs, not demographics. Describe primary and secondary segments with size estimates.
  6. Value Proposition(s) -- Map customer jobs addressed, gains provided, and pain points eliminated. Show competitive differentiation using frameworks like Value Curve analysis.
  7. Solution -- Feature descriptions, UX/prototypes, wireframes, user flows, and technology details when relevant. Include out-of-scope items explicitly. Document assumptions. Enumerate at least 5 edge cases.
  8. Release -- Phased rollout plan using relative timeframes (not exact dates). Define MVP vs. future iterations, feature flags, rollback criteria, and review checkpoints.

For lightweight PRDs, sections 2, 3, and 8 can be condensed to 2-3 sentences each.

Frameworks & Best Practices

  • Problem before solution. Spend 40% of the document on sections 1-5 (the "why") before touching section 7 (the "what"). A PRD that jumps to the solution is a spec, not a PRD.
  • One objective, not five. A PRD with multiple objectives is multiple PRDs. Split them. Each PRD should have a single primary metric it moves.
  • Market segments defined by needs. Describe who this is for based on the problems they face and jobs they hire the product to do, not by demographics or firmographics alone.
  • Value Proposition clarity. For each segment, explicitly state the customer jobs addressed, gains provided, and pains eliminated. Use the Value Curve to show where you differentiate from competitors.
  • Data-driven specificity. Replace vague language with specific numbers. "Improve retention" is not a metric; "Increase D7 retention from 25% to 35% within 8 weeks of launch" is.
  • Scope creep guard. Explicitly list what is NOT in scope. Revisit the out-of-scope list when stakeholders propose additions.
  • Relative timeframes over dates. Use phases and relative windows rather than exact calendar dates. This prevents false precision and allows flexibility.
  • Assumption tracking. List the top 3 assumptions underlying the PRD. For each, state supporting evidence and what would invalidate it.
  • Audience awareness. Engineers need technical constraints and edge cases. Designers need user flows and personas. Executives need the summary and metrics. Write for all three in a single document.
  • Living document. State the PRD version and last-updated date. PRDs that never change were never read.
  • Lightweight PRD triggers: pre-PMF exploration, hackathon projects, internal tools, experiments with <2 week timelines.
  • Full PRD triggers: cross-team initiatives, features with external dependencies, anything touching payments or compliance.

Related Skills

  • roadmap-planning -- Chain after writing PRDs to slot the initiative into the broader roadmap with dependencies and timelines.
  • mvp-scoping -- Chain before writing a PRD to determine what to include in v1 vs. defer to later releases.
  • user-research-synthesis -- Chain before writing a PRD to ground the Background and Market Segments sections in real customer data.

Examples

Example 1: Lightweight PRD prompt

User: "We need a PRD for adding Slack notifications to our project management tool."

Good output excerpt:

Summary: Add configurable Slack notifications so that teams using ProjectFlow are alerted to task updates, mentions, and deadline changes without leaving their primary communication tool. This addresses the #1 feature request from our Q3 customer survey (38% of respondents). Objective: Enable team leads on paid plans to receive real-time project updates in Slack, resulting in a 20% reduction in average response time to task assignments within 6 weeks of launch. Market Segment: Teams of 10+ coordinating across tools, whose primary pain is context-switching between project management and communication platforms. Out of scope: Microsoft Teams integration, custom notification templates, Slack bot commands.

Example 2: Full PRD prompt

User: "Write a full PRD for our new self-serve onboarding flow. We're losing 60% of signups before they complete setup."

Good output excerpt:

Background: Current onboarding requires 7 steps and takes an average of 12 minutes. Hotjar recordings show 45% of users abandon at the "connect data source" step. Competitor X reduced their onboarding to 3 steps in Q2 and reported a 2x improvement in activation. Our support team handles 30+ onboarding tickets per week, costing approximately $4,500/month. Value Proposition: Eliminate the "connect data source" friction by offering a sample dataset that lets users experience core value before committing to integration. Differentiated from Competitor X which still requires immediate data connection. Release: - Phase 1: Internal dogfood with the team (2 weeks) - Phase 2: 10% of new signups via feature flag - Phase 3: 50% rollout if activation rate > 45% - Phase 4: GA if no P0 bugs and support ticket volume decreases - Rollback trigger: activation rate drops below current 40% baseline

平台分布

Codex

35.68%
按下载量换算105

Claude

30.55%
按下载量换算90

Cursor

19.9%
按下载量换算58

Gemini CLI

8.69%
按下载量换算25

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