Token导航 LogoToken导航

Agent Skill · 研究检索

spec(规格说明)

spec 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 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,权限为需要联网;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 45,286。

总安装 857周安装 35GitHub Stars 2,473下载 277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(规格说明)
来源仓库:https://github.com/garden-co/jazz
仓库路径:skills/spec
安装命令:
npx skills add https://github.com/garden-co/jazz --skill spec
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/garden-co/jazz --skill spec

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 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。适用于关键词搜索、任务场景匹配或来源线索梳理等研究检索场景。通过 npx skills add 命令从指定 GitHub 仓库安装并使用。

安装前需确认权限范围、维护状态及是否触发联网或文件操作。建议结合原始 README 核验具体用法和功能边界。

来源仓库内路径:skills/spec

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/garden-co/jazz --skill spec;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

    它需要需要联网,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

spec 怎么安装?

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

spec 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 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

spec 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 用的人多吗?

公开数据是累计安装 857 次,近一周安装 35 次,下载 277 次,GitHub 2,473 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 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

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。
  • 来源仓库有 2,473 星,维护活跃度相对可查。

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。

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

规格说明 · 本页

8572,473需要联网未明确未通过

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

Spec Driven Development Skill

Use this skill when the user asks to implement a feature. This workflow ensures proper planning and approval before any code is written.

Workflow

Follow these steps in order. Do not skip steps. Always ask for explicit approval before moving to the next step.

Step 1: Create Feature Directory

Create a feature directory under .specs/{feature_name} using kebab-case for the name.

Step 2 (Optional): Create Requirements Document

This step is opt-in and must be explicitly requested by the user.

If the user requests requirements, create requirements.md in the feature directory with:

  • Introduction - Brief context and purpose of the feature
  • User Stories & Acceptance Criteria - Written in EARS (Easy Approach to Requirements Syntax) style

Example EARS patterns:

  • Ubiquitous: "The [system] shall [action]"
  • Event-driven: "When [event], the [system] shall [action]"
  • State-driven: "While [state], the [system] shall [action]"
  • Optional: "Where [condition], the [system] shall [action]"
  • Unwanted behavior: "If [condition], then the [system] shall [action]"

Example structure:

# Requirements

## Introduction

[Brief context about what this feature addresses and why it's needed]

## User Stories & Acceptance Criteria

### US-1: [User Story Title]

**As a** [role], **I want** [goal], **so that** [benefit].

**Acceptance Criteria:**
- When [user action], the system shall [expected behavior]
- While [state], the system shall [maintain condition]
- Where [optional condition], the system shall [handle appropriately]

After creating the requirements document, refine it:

  1. Scan for missing requirements — Review the document for gaps such as missing edge cases, undefined behavior, unspecified error handling, unclear scope boundaries, missing performance constraints, or unaddressed user roles/permissions.
  2. Identify ambiguities — Flag any requirements that could be interpreted in multiple ways, have vague language (e.g., "fast", "simple", "flexible"), or lack concrete acceptance criteria.
  3. Ask clarifying questions — Present the user with a clear list of questions covering the missing and ambiguous areas. Group them logically and explain why each question matters.

Do not proceed until all critical ambiguities are resolved. Minor open questions can be noted as assumptions in the design document.

If requirements are created, get approval before proceeding to the design document.

Step 3: Create Design Document

Create design.md in the feature directory with:

  • Overview - High-level description of the solution
  • Architecture / Components - System structure and component interactions
  • Data Models - Schemas, types, and data structure
  • Testing Strategy - Approach to testing the feature

Show the code snippets of the core parts of the implementation in the design.

We prioritize integration testing, and show a couple of test snippets as example of testing strategy.

After creating the design document, refine it:

  1. Scan for missing requirements — Check whether the design covers all stated requirements and user stories. Identify any requirements that were dropped, under-specified, or only partially addressed.
  2. Identify ambiguities — Flag design decisions that are vague or leave open questions about behavior, data flow, or component responsibilities.
  3. Ask clarifying questions — Present the user with questions about any gaps or ambiguities discovered. Explain how each gap could affect implementation.

Do not proceed until all critical gaps are resolved. Minor open questions can be noted as assumptions.

Step 4: Design Approval Gate

Ask the user: "Does the design look good? If so, we can move on to the implementation plan."

Wait for explicit approval before proceeding.

Step 5: Create Tasks Document

Once design is approved, create tasks.md in the feature directory with:

  • Numbered checklist of coding tasks
  • Each task should reference specific design components
  • Include only coding tasks - no deployment, documentation, or other non-coding tasks

Example structure:

# Implementation Tasks

## Tasks

- [ ] 1. Create data model for [entity]
- [ ] 2. Add API endpoint for [action]
- [ ] 3. Implement validation logic
- [ ] 4. Add unit tests for [component]
- [ ] 5. Add integration tests for [feature]

Step 6: Tasks Approval Gate

Ask the user: "Do the tasks look good?"

Wait for explicit approval.

Step 7: Stop

Do not implement any code. The workflow ends here. Implementation should be a separate activity initiated by the user.

Important Rules

  1. Never skip steps - Each step builds on the previous one
  2. Always get approval - Do not proceed without explicit user confirmation
  3. No implementation - This workflow is for planning only
  4. Kebab-case naming - Feature directories use kebab-case (e.g., user-authentication)

平台分布

Codex

34.26%
按下载量换算95

Claude

28.75%
按下载量换算80

Cursor

17.09%
按下载量换算47

Gemini CLI

9.97%
按下载量换算28

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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