Token导航 LogoToken导航

Agent Skill · 开发

specs-setup(规格设置)

specs-setup 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。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。

specs-setup 是开发类 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,按安装量在开发类的 21,746 个 Skill 里排第 21,023。

总安装 196周安装 8GitHub Stars 2下载 63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:specs-setup(规格设置)
来源仓库:https://github.com/b12consulting/skills
仓库路径:skills/specs-setup
安装命令:
npx skills add https://github.com/b12consulting/skills --skill specs-setup
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/b12consulting/skills --skill specs-setup

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

specs-setup 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息。适用于开发初始化或配置阶段的规格管理与协作支持。使用 npx skills add 命令从指定仓库安装并使用。安装前需评估权限、项目维护情况及是否触发敏感操作。

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/specs-setup

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/b12consulting/skills --skill specs-setup;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

specs-setup 怎么安装?

在终端执行 npx skills add https://github.com/b12consulting/skills --skill specs-setup。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

specs-setup 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。

specs-setup 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。

specs-setup 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。

specs-setup 用的人多吗?

公开数据是累计安装 196 次,近一周安装 8 次,下载 63 次,GitHub 2 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。

specs-setup 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。
  • 需要处理开发类任务、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。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
specs-setup

规格设置 · 本页

1962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

Specs Setup

Initialize the spec-driven methodology for a project. Run this when specs/ doesn't exist or is missing critical files.

Skill Dependencies

This skill is part of a set of three skills designed to work together:

  • spec-driven — Methodology reference (structure, formats, rules)
  • specs-setup (this skill) — Initialize specs/ for a new project
  • specs-tickets — Create and execute tickets through their lifecycle
  • specs-review — Audit specs health, consistency, and drift

If any of these skills are missing from the project, instruct the user to install them before proceeding:

npx skills add b12consulting/skills --skill <missing_skill>

Always load the spec-driven skill first for the full methodology reference. Load templates when creating documents.

Prerequisites

Before setting up specs, check if the project has coding standards defined:

  • .github/copilot-instructions.md or .github/instructions/*.instructions.md
  • CLAUDE.md
  • Or equivalent

If no coding standards exist, prompt the user to create them before proceeding. Specs define WHAT to build; coding standards define HOW to build it. Both are needed before any implementation work begins.

Procedure

1. Assess Current State

Check which of these exist:

  • specs/ directory
  • specs/README.md
  • specs/Vision.md
  • specs/PRD.md
  • specs/Goals.md
  • specs/Architecture/ directory
  • specs/Architecture/README.md
  • specs/Glossary.md
  • specs/Changelog.md
  • specs/decisions/ directory
  • specs/tickets/ directory

Report what's missing and confirm with the user before creating anything.

2. Create Missing Structure

Create any missing directories and files using the templates from templates.md.

Create in this order:

  1. specs/ directory
  2. specs/decisions/ directory (add a .gitkeep if empty)
  3. specs/tickets/ directory (add a .gitkeep if empty)
  4. specs/Architecture/ directory
  5. specs/Vision.md
  6. specs/PRD.md
  7. specs/Goals.md
  8. specs/Architecture/README.md
  9. specs/Glossary.md
  10. specs/Changelog.md
  11. specs/README.md (last, because it links to everything above)

3. Populate Vision.md

Interview the user to fill in the vision. Ask about:

  1. Vision statement: What is this project? Why does it exist? What future are we building toward?
  2. Problem statement: What problem does it solve? Who feels this pain? What is the impact?
  3. Target audience: Who are the primary users or personas?

Write Vision.md based on the user's answers. Present it for review and confirmation.

If the project has existing documentation or a README, use it as input — but always confirm with the user rather than assuming.

4. Populate PRD.md

Interview the user to fill in the PRD. Ask about:

  1. Functional requirements: What are the key things the system must do?
  2. Non-functional requirements: Performance, security, scalability, accessibility needs?
  3. Scope: What's explicitly in scope? What's explicitly out of scope?
  4. Assumptions & constraints: What are we assuming? What limits us?

Write PRD.md based on the user's answers. Present it for review and confirmation.

5. Populate Goals.md

Interview the user to fill in the goals. Ask about:

  1. Success metrics: How will we know it's successful? What are the quantitative and qualitative indicators?
  2. Milestones: What are the key milestones and their target dates?

Write Goals.md based on the user's answers. Present it for review and confirmation.

6. Populate Architecture

If the project has existing code, analyze it and draft the architecture documentation:

  1. System overview: What does the system do at a high level?
  2. Key components: What are the major parts and their responsibilities?
  3. Technology stack: What technologies are used and why?
  4. Key constraints: What are the important architectural constraints and trade-offs?

If the project is new (no code yet), work with the user to define the target architecture.

Present Architecture/README.md for review and confirmation.

Remember: keep it high-level. The architecture entry point should give someone a clear mental model of the system in under 5 minutes of reading. Split into sub-documents only when a section exceeds ~200 lines.

7. Populate Glossary

Scan the Vision, PRD, and Architecture for domain-specific terms. Draft definitions and ask the user to confirm them. Even a small initial glossary (5-10 terms) is valuable — it can grow over time.

8. Create Instructions File

Create a .github/instructions/specs.instructions.md file to ensure agents automatically load the spec-driven methodology when working with specs:

---
applyTo: "specs/**"
---

This project uses the spec-driven methodology. Load the `spec-driven` skill before making any changes to files in the specs/ folder.

If the .github/instructions/ directory doesn't exist, create it.

9. Finalize README

Update specs/README.md with:

  • The project name and one-line description
  • Current status (likely "Setting up" or "No active tickets")
  • Working navigation links to all created documents

10. Suggest Commit

Suggest the user commits the initial specs:

docs: initialize spec-driven methodology

Incremental Setup

If specs/ already exists but is incomplete, only create the missing pieces. Do not overwrite existing documents — they may contain work that shouldn't be lost. Instead, flag any inconsistencies or gaps and let the user decide how to resolve them.

平台分布

Codex

34.38%
按下载量换算22

Claude

34%
按下载量换算21

Cursor

17.41%
按下载量换算11

Gemini CLI

9.64%
按下载量换算6

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