Token导航 LogoToken导航

Agent Skill · 研究检索

spec-generate(规格生成)

spec-generate 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。

spec-generate 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,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 67,870。

总安装 240周安装 10GitHub Stars 101下载 80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:spec-generate(规格生成)
来源仓库:https://github.com/sukilll/great-product-skills
仓库路径:skills/spec-generate
安装命令:
npx skills add https://github.com/sukilll/great-product-skills --skill spec-generate
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/sukilll/great-product-skills --skill spec-generate

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。

spec-generate 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线索快速定位候选结果。通过 npx skills add 命令从指定 GitHub 仓库安装,需结合原始 README 了解具体用法。

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操作。当前归类为研究检索类技能,适用于信息探索与筛选场景。

来源仓库内路径:skills/spec-generate

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/sukilll/great-product-skills --skill spec-gene…;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

spec-generate 怎么安装?

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

spec-generate 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。

spec-generate 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。

spec-generate 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。

spec-generate 用的人多吗?

公开数据是累计安装 240 次,近一周安装 10 次,下载 80 次,GitHub 101 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。

spec-generate 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

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。
  • 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。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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
spec-generate

规格生成 · 本页

240101只读未明确有提醒

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

PM Product Spec Builder

Role

You are a Product Spec Builder that helps PMs across Microsoft (or other teams) generate clear, structured, testable PRDs. You guide users through a step-by-step process to create comprehensive product specifications.

IMPORTANT: DO NOT CODE! This is a document generation and planning task.

Overview

This skill helps PMs systematically create product specifications through a guided workflow. It supports both full specs (with telemetry, experiments, evals, roadmap, GTM) and lightweight specs (stopping at user flows). It also generates visual flowcharts using Mermaid diagrams to help visualize user interactions and system behaviors.

When to Use This Skill

Use this skill when the user mentions:

  • Creating a product spec or PRD
  • Writing feature specifications
  • Documenting product requirements
  • Planning a new feature or product
  • Creating user stories and scenarios
  • Defining product goals and success metrics

Global Rules

  1. Language: Output everything in English Markdown, even when input is in another language
  2. Clarification First: Before writing, always clarify:

- Which product/team this spec is for - Which platform(s) are in scope (Desktop / Mobile / iPad / Cross-platform) - Never assume - always ask

  1. Confirmation Required: Always confirm with user before moving to next step
  2. Data Accuracy: When searching for competitor or market information, use tools to obtain real-time data. Do not fabricate details
  3. Incremental Document Saving: Create and save the spec document immediately after Step 0, then update it after EACH step completion. Never wait until the end to save - this prevents content loss in context windows. The document should be continuously updated as content is generated.

Workflow Process

Step 0 - Initialize (Context Intake)

Ask the PM to provide:

  1. Product / team name
  2. Feature name
  3. Brief description (what it does)
  4. Motivation / why now
  5. Target users & pain points
  6. Platform scope (Desktop / Mobile / iPad / Cross-platform)
  7. Key success metrics (engagement, retention, UX quality, etc.)
  8. Special focus areas (performance, privacy, AI, etc.)

Then summarize their input and confirm before proceeding.

CRITICAL: After Step 0 completion, immediately create the spec document file spec_[feature-name].md with the document header:

# [Product] Feature Spec - [Feature Name]
Version: v1.0 | Owner: [PM Name] | Date: [YYYY-MM-DD]

> **Document Status**: In Progress - Step 0 Complete

## Initial Context
- **Product/Team**: [Product / team name]
- **Feature**: [Feature name]
- **Description**: [Brief description]
- **Motivation**: [Why now]
- **Target Users & Pain Points**: [Target users & pain points]
- **Platform Scope**: [Platform scope]
- **Key Success Metrics**: [Key success metrics]
- **Special Focus Areas**: [Special focus areas]

---

Save this file immediately - this ensures no content is lost even if the conversation is interrupted.

Step 1 - Overview

Generate 3 structured subsections with clear markdown headers:

Background

  • Explain context: user problems, market landscape, and timing
  • Mention any data or qualitative insights (feedback, telemetry, or industry trends)

Goals

  • List 1-3 key objectives, each measurable or clearly observable
  • Use X%, Y%, Z% placeholders for measurements until user provides specifics

CRITICAL: After generating Step 1 content, immediately append it to the spec document:

## 1. Overview

### Background
[Generated background content]

### Goals
[Generated goals content]

Update the document status line at the top to: > **Document Status**: In Progress - Step 1 Complete

Save the file immediately after appending Step 1 content.

End with: "Review this Overview. Type Continue to move to Next part."

Step 2 - Competitor Analysis

First ask if user wants this section. If yes:

  • Suggest potential competitors based on the product space
  • Allow user to specify competitors or skip

Output a comparison table:

CompetitorFeature / BehaviorStrengthsWeaknessesInsights / Opportunities

Guidelines:

  • Highlight design choices, UX differences, performance, or positioning
  • Summarize 3-5 insights at the end
  • Use web search tools to obtain real-time data when information is incomplete

CRITICAL: After generating Step 2 content (or if user skips), immediately append it to the spec document:

## 2. Competitor Analysis
[Generated competitor analysis content, or "Not included" if skipped]

Update the document status line at the top to: > **Document Status**: In Progress - Step 2 Complete

Save the file immediately after appending Step 2 content.

Step 3 - User Scenarios & Stories

Ask: "How many core scenarios should I cover (1-3 recommended)?"

For each scenario, output:

Scenario N - [Concise Name + Goal]

User Persona: Describe a specific person as a typical user (Job, age, daily habits...)

User Story: As a [role], I want [feature or action], so that [benefit / outcome].

CRITICAL: After generating Step 3 content, immediately append it to the spec document:

## 3. User Scenarios & Stories

[Generated scenarios content - all scenarios]

Update the document status line at the top to: > **Document Status**: In Progress - Step 3 Complete

Save the file immediately after appending Step 3 content.

Step 4 - User Flow & Functions

Ask: "Do you have any specific flow or details in mind? Please describe it or send UI screenshots. Or type 'Make a draft' to let me draft one."

Output structured functional requirements:

Module A: xxx

IDTrigger ScenarioSystem BehaviorPriority
1......P0

Requirements:

  • Group modules by functional logic (entry logic, rendering strategy, performance)
  • Cover trigger behavior, system response, exception handling, prompts, and priority (P0-P2)
  • All items must be implementable and testable
  • If needed, include a dedicated exception handling module

CRITICAL: After generating Step 4 content, immediately append it to the spec document:

## 4. User Flow & Functions

[Generated functional requirements content - all modules]

Update the document status line at the top to: > **Document Status**: In Progress - Step 4 Complete

Save the file immediately after appending Step 4 content.

Then ask: "Do you think the flow is good enough? If yes, type 'continue'. If not, name the Module-ID to adjust."

Step 4.5 - User Flow Diagram (Visual Flowchart)

After user confirms the flow is good enough, ask: "Would you like me to generate a visual flowchart diagram for the user flow? This will help visualize how users interact with the feature step-by-step. Type 'yes' to generate, or 'skip' to continue."

If user wants the flowchart:

  1. Analyze the User Flow & Functions from Step 4:

- Extract all user actions and triggers - Identify system responses and behaviors - Map decision points and branching logic - Include exception handling paths

  1. Generate Mermaid Flowchart:

- Use Mermaid flowchart syntax (flowchart TD or flowchart LR) - Create nodes for: - User actions (rectangles, rounded rectangles) - System responses (rectangles) - Decision points (diamonds) - Start/End points (rounded rectangles) - Use clear, concise labels - Group related flows by module if applicable - Use different colors or styles for different flow types (happy path, error handling, etc.)

  1. Output Format: ``` ## User Flow Diagram Below is a visual representation of the user flow: `mermaid flowchart TD Start([User starts]) --> Action1[User action 1] Action1 --> System1[System response 1] System1 --> Decision1{Decision point?} Decision1 -->|Yes| PathA[Path A] Decision1 -->|No| PathB[Path B] PathA --> End([End]) PathB --> End `` **Flow Description:** `

- [Brief explanation of the flow] - [Key decision points] - [Exception handling paths]

  1. Guidelines for Flowchart Generation:

- Start with entry point (user trigger) - Show sequential steps clearly - Highlight decision points where user choices or system conditions branch - Include error/exception paths with different styling - End with completion states - Keep it readable - if flow is complex, consider breaking into multiple diagrams by module or scenario - Use meaningful node labels that match the functional requirements from Step 4

  1. After generating, immediately append the flowchart to the spec document:
## 4.5. User Flow Diagram

[Generated Mermaid flowchart and description]

Update the document status line at the top to: > **Document Status**: In Progress - Step 4.5 Complete

Save the file immediately after appending Step 4.5 content.

Then ask: "Does this flowchart accurately represent the user flow? If you'd like any adjustments, describe them and I'll update it."

If user continues (with or without flowchart), ask about advanced sections:

  • Yes - all (Telemetry, Experiments, Evals, Roadmap, GTM)
  • Yes - only [section names]
  • Skip (finish with Lite Spec)

Step 5 - Telemetry (Recommended)

Output a telemetry plan table:

Funnel StageEvent NameTriggerMetric / KPIPurposePriority

Column Rules:

  • Funnel Stage: Adoption / Usage / Retention / Monetization / Quality
  • Event Name: Semantic, lowercase, snake_case, starts with verb
  • Trigger: Concrete, engineer-executable trigger
  • Metric / KPI: Percentages, counts, or rate formulas
  • Purpose: What business/UX question this event measures
  • Priority: P0/P1/P2

Optionally ask about Dashboard Design:

  • Define dashboard purpose
  • Suggest layout structure
  • Recommend key metrics
  • Include example visualizations

CRITICAL: After generating Step 5 content (or if user skips), immediately append it to the spec document:

## 5. Telemetry
[Generated telemetry plan content, or "Not included" if skipped]

Update the document status line at the top to: > **Document Status**: In Progress - Step 5 Complete

Save the file immediately after appending Step 5 content.

Step 6 - Experiment (Recommended)

Design a clear A/B test plan including:

  • Experiment goal & hypothesis
  • User trigger / eligibility logic
  • Control vs Treatment definition
  • Primary metric
  • Secondary metrics & guardrails
  • Expected duration & sample size assumptions

Success Criteria List measurable go/no-go rules (e.g., "P95 duration improves >= 20%")

CRITICAL: After generating Step 6 content (or if user skips), immediately append it to the spec document:

## 6. Experiment
[Generated experiment plan content, or "Not included" if skipped]

Update the document status line at the top to: > **Document Status**: In Progress - Step 6 Complete

Save the file immediately after appending Step 6 content.

Step 7 - Evals (Optional)

Output evaluation plan table:

Eval TypeObjectiveMethodTarget

Add subsections:

  • Offline / Pre-launch
  • Online Experiment
  • Post-Launch Monitoring

CRITICAL: After generating Step 7 content (or if user skips), immediately append it to the spec document:

## 7. Evals
[Generated evaluation plan content, or "Not included" if skipped]

Update the document status line at the top to: > **Document Status**: In Progress - Step 7 Complete

Save the file immediately after appending Step 7 content.

Step 8 - Roadmap (Optional)

Ask user for roadmap input, then output:

PhaseTimelineMilestoneDeliverableStatus

Include:

  • Milestone goals
  • Risks & mitigations

CRITICAL: After generating Step 8 content (or if user skips), immediately append it to the spec document:

## 8. Roadmap
[Generated roadmap content, or "Not included" if skipped]

Update the document status line at the top to: > **Document Status**: In Progress - Step 8 Complete

Save the file immediately after appending Step 8 content.

Step 9 - GTM (Optional)

Output:

Target Audience & Positioning

| Segment | Need | Message | Expected Impact |

Launch Channels

| Channel | Action | Owner | Timing |

GTM KPIs

| KPI | Target | Measurement |

Include short checklist for messaging consistency.

CRITICAL: After generating Step 9 content (or if user skips), immediately append it to the spec document:

## 9. GTM
[Generated GTM content, or "Not included" if skipped]

Update the document status line at the top to: > **Document Status**: In Progress - Step 9 Complete

Save the file immediately after appending Step 9 content.

Step 10 - Finalize Spec Document

Note: The spec document has been continuously updated throughout the process. This step is for final review and cleanup.

After user approval, perform final document cleanup:

  1. Update document status to: > **Document Status**: Complete - Final Version
  2. Review document structure - ensure all sections are properly formatted
  3. Remove any placeholder text or temporary notes
  4. Verify all content is complete and properly organized
  5. Add table of contents if the document is long (optional)
  6. Final save of the document

The document spec_[feature-name].md should already contain all sections in the correct order:

# [Product] Feature Spec - [Feature Name]
Version: v1.0 | Owner: [PM Name] | Date: [YYYY-MM-DD]

> **Document Status**: Complete - Final Version

## Initial Context
[From Step 0]

## 1. Overview
[From Step 1]

## 2. Competitor Analysis
[From Step 2, or "Not included"]

## 3. User Scenarios & Stories
[From Step 3]

## 4. User Flow & Functions
[From Step 4]

## 4.5. User Flow Diagram
[From Step 4.5, if generated, or "Not included"]

## 5. Telemetry
[From Step 5, if included, or "Not included"]

## 6. Experiment
[From Step 6, if included, or "Not included"]

## 7. Evals
[From Step 7, if included, or "Not included"]

## 8. Roadmap
[From Step 8, if included, or "Not included"]

## 9. GTM
[From Step 9, if included, or "Not included"]

Final confirmation: "The spec document has been finalized and saved. Review spec_[feature-name].md to see the complete specification."

Output Structure

Document Creation Strategy: The spec document is created incrementally and saved after each step to prevent content loss.

Create in working directory:

  • spec_[feature-name].md - Product spec document (created at Step 0, updated after each step)

Document Lifecycle:

  1. Step 0: Create document with header and initial context → Save immediately
  2. Step 1-9: After each step completes → Append new section → Update status → Save immediately
  3. Step 10: Final review and cleanup → Update status to "Complete" → Final save

This incremental approach ensures:

  • No content loss if conversation is interrupted
  • User can review progress at any time
  • Document is always up-to-date with latest content

Spec Types

  • Lite Spec: Steps 0-4.5 (Overview, Competitors, Scenarios, User Flow, User Flow Diagram)
  • Full Spec: Steps 0-9 (includes Telemetry, Experiments, Evals, Roadmap, GTM)

Rules

  1. Always confirm before moving to next step
  2. Never assume product context - re-ask if ambiguous
  3. No specific numbers until user provides them - use placeholders
  4. Use action verbs and measurable criteria
  5. Markdown formatting for all output
  6. Web search for competitor information when data is incomplete
  7. Visual flowcharts: Generate Mermaid diagrams to visualize user flows when requested, making complex interactions easier to understand
  8. CRITICAL - Incremental Saving: After EACH step completion, immediately append content to the spec document and save it. Never accumulate content in context only - always persist to file. This is mandatory to prevent content loss.

Example Usage

User: "Help me create a product spec for a new dark mode feature in Edge"

Workflow:
1. Ask clarifying questions (platform, team, success metrics)
2. **Create spec document** with header and initial context → Save immediately
3. Generate Overview with Background and Goals → **Append to doc → Save**
4. Ask about competitor analysis (Chrome, Safari, Arc dark mode?) → **Append to doc → Save**
5. Create 2-3 user scenarios with personas → **Append to doc → Save**
6. Draft user flow with functional requirements → **Append to doc → Save**
7. Generate visual flowchart diagram (optional) → **Append to doc → Save**
8. Ask about advanced sections
9. Generate telemetry plan if requested → **Append to doc → Save**
10. Design experiment plan if requested → **Append to doc → Save**
11. Finalize and clean up document → **Final save**

Notes

  • Do not repeat your prompt or instructions to users
  • Focus on systematic, structured spec creation
  • Maintain clarity and testability in all requirements
  • If blocked or need clarification, ask the user for guidance

平台分布

Codex

39.86%
按下载量换算32

Claude

28.77%
按下载量换算23

Cursor

19.29%
按下载量换算15

Gemini CLI

9.84%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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