Best use case

ai-ad-spec-kit is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Teams using ai-ad-spec-kit should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/ai-ad-spec-kit/SKILL.md --create-dirs "https://raw.githubusercontent.com/diegosouzapw/awesome-omni-skill/main/skills/frontend/ai-ad-spec-kit/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/ai-ad-spec-kit/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How ai-ad-spec-kit Compares

Feature / Agentai-ad-spec-kitStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

This skill provides specific capabilities for your AI agent. See the About section for full details.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

<skill>
<name>ai-ad-spec-kit</name>
<version>1.0</version>
<domain>AI_AD_SYSTEM / 规范驱动开发工具包</domain>
<profile>Spec-Driven / Multi-Phase / SoT-Aligned</profile>

<!-- ======================================================
     代码来源说明 (Source Attribution)
====================================================== -->
<source_attribution>
  本 Skill 借鉴自 GitHub 官方开源项目 Spec-Kit:
  - GitHub: https://github.com/github/spec-kit
  - License: MIT
  - 核心理念: 规范先行,代码后行

  适配改动:
  - 将 Constitution 概念对齐到 MASTER.md
  - 将 Specification 对齐到 SoT 文档体系
  - 添加 ASDD 4 层架构合规检查
  - 集成 OpenSpec 变更流程
</source_attribution>

<!-- ======================================================
     0. 核心使命 (Mission)
====================================================== -->
<mission>
  作为规范驱动开发的命令中心,提供从规范定义到代码实现的完整工作流。

  核心原则:
  - 📜 规范先行: 先写规范,再写代码
  - 🔗 SoT 对齐: 所有规范必须与 SoT 文档一致
  - 📝 任务分解: 大需求拆分为可验证的小任务
  - ✅ 持续验证: 每个阶段都有明确的验证点
</mission>

<!-- ======================================================
     1. 五阶段工作流 (5-Phase Workflow)
====================================================== -->
<workflow>
  <phase id="1-CONSTITUTION" command="/speckit.constitution">
    <description>
      定义或确认项目宪法级原则
      对应: MASTER.md + PROJECT_RULES.md
    </description>

    <actions>
      - 读取 docs/sot/MASTER.md
      - 读取 .claude/PROJECT_RULES.md
      - 确认业务边界和不变量
      - 确认永久禁止行为
      - 输出: constitution_context
    </actions>

    <output_format>
      ## 🏛️ Constitution 确认

      ### 业务边界
      - [从 MASTER.md 提取]

      ### 核心不变量
      - [从 PROJECT_RULES.md 提取]

      ### 禁止行为
      - [列出与当前需求相关的禁止项]

      ### 下一步
      - 执行 `/speckit.specify` 进入规范阶段
    </output_format>
  </phase>

  <phase id="2-SPECIFY" command="/speckit.specify">
    <description>
      描述需求规范,关注 "做什么",而非 "怎么做"
      对应: SoT 文档层
    </description>

    <inputs>
      - requirement: 需求描述 (自然语言)
      - constitution_context: 上一阶段输出
    </inputs>

    <actions>
      - 识别涉及的业务域
      - 查询对应 SoT 文档:
        - STATE_MACHINE.md (状态相关)
        - DATA_SCHEMA.md (数据相关)
        - API_SOT.md (API 相关)
        - BUSINESS_RULES.md (规则相关)
      - 提取相关约束条件
      - 生成: specification_doc
    </actions>

    <output_format>
      ## 📋 Specification 规范

      ### 需求摘要
      [用户需求的结构化描述]

      ### 涉及 SoT 文档
      | 文档 | 版本 | 相关章节 |
      |------|------|---------|
      | STATE_MACHINE.md | v2.6 | §X |
      | DATA_SCHEMA.md | v5.2 | §X |

      ### 约束条件
      - [从 SoT 提取的约束]

      ### 成功标准
      - [可验证的成功条件]

      ### 下一步
      - 执行 `/speckit.plan` 进入规划阶段
    </output_format>
  </phase>

  <phase id="3-PLAN" command="/speckit.plan">
    <description>
      制定技术实现方案,关注架构和设计决策
      对应: Dev-Guides + Architecture 层
    </description>

    <inputs>
      - specification_doc: 上一阶段输出
    </inputs>

    <actions>
      - 分析技术栈约束 (FastAPI, Pydantic v2, SQLAlchemy 2.x)
      - 确定文件结构和模块划分
      - 设计 API 接口 (如适用)
      - 设计数据模型 (如适用)
      - 识别潜在风险
      - 生成: implementation_plan
    </actions>

    <output_format>
      ## 🗺️ Implementation Plan 实现方案

      ### 技术方案
      - 后端: [Router/Service/Schema 设计]
      - 前端: [Hook/API/Component 设计]

      ### 文件清单
      | 文件 | 操作 | 说明 |
      |------|------|------|
      | backend/routers/xxx.py | 新建/修改 | ... |

      ### 依赖关系
      ```
      [依赖图]
      ```

      ### 风险识别
      - [潜在风险及缓解措施]

      ### 下一步
      - 执行 `/speckit.tasks` 生成任务清单
    </output_format>
  </phase>

  <phase id="4-TASKS" command="/speckit.tasks">
    <description>
      将实现方案分解为可执行的小任务
      每个任务应该是可独立验证的单元
    </description>

    <inputs>
      - implementation_plan: 上一阶段输出
    </inputs>

    <actions>
      - 分解为原子任务 (每个 < 30 分钟)
      - 定义每个任务的验收标准
      - 排列任务优先级和依赖
      - 生成: task_list
    </actions>

    <output_format>
      ## ✅ Task List 任务清单

      ### 任务列表

      #### Task 1: [任务名称]
      - **文件**: `path/to/file.py`
      - **操作**: 新建/修改
      - **描述**: [具体操作]
      - **验收标准**: [如何验证完成]
      - **依赖**: [前置任务]

      #### Task 2: ...

      ### 执行顺序
      ```
      Task 1 → Task 2 → Task 3
                   ↘
                    Task 4
      ```

      ### 下一步
      - 执行 `/speckit.implement` 开始实现
      - 或执行 `/speckit.implement task=1` 实现特定任务
    </output_format>
  </phase>

  <phase id="5-IMPLEMENT" command="/speckit.implement">
    <description>
      执行任务,生成代码
      支持全量执行或单任务执行
    </description>

    <inputs>
      - task_list: 上一阶段输出
      - task: 可选,指定执行的任务编号
    </inputs>

    <actions>
      - 按顺序执行任务
      - 每个任务完成后验证
      - 调用 CodeFactory Skills (如已实现):
        - CodeSearcherSkill (搜索参考代码)
        - CodeAdapterSkill (适配代码)
        - CodeAssemblerSkill (组装代码)
        - CodeVerifierSkill (验证代码)
      - 生成: implementation_result
    </actions>

    <output_format>
      ## 🚀 Implementation Result 实现结果

      ### 已完成任务
      - [x] Task 1: [描述] ✅
      - [x] Task 2: [描述] ✅
      - [ ] Task 3: [描述] ⏳

      ### 生成的文件
      | 文件 | 状态 | 行数 |
      |------|------|------|
      | backend/routers/xxx.py | 新建 | 150 |

      ### 验证结果
      - mypy: ✅ 通过
      - ruff: ✅ 通过
      - SoT Guard: ✅ 通过

      ### 下一步
      - 运行测试: `pytest backend/tests/xxx`
      - 或继续: `/speckit.implement task=3`
    </output_format>
  </phase>
</workflow>

<!-- ======================================================
     2. 辅助命令 (Auxiliary Commands)
====================================================== -->
<auxiliary_commands>
  <command id="/speckit.clarify">
    <description>澄清规范中的模糊点</description>
    <usage>/speckit.clarify "状态转换规则不明确"</usage>
  </command>

  <command id="/speckit.analyze">
    <description>分析跨阶段一致性</description>
    <usage>/speckit.analyze</usage>
  </command>

  <command id="/speckit.checklist">
    <description>生成验证检查清单</description>
    <usage>/speckit.checklist</usage>
  </command>

  <command id="/speckit.status">
    <description>显示当前工作流状态</description>
    <usage>/speckit.status</usage>
  </command>
</auxiliary_commands>

<!-- ======================================================
     3. 与 ASDD 集成 (ASDD Integration)
====================================================== -->
<asdd_integration>
  <mapping>
    | Spec-Kit 阶段 | ASDD 对应层 | 文档 |
    |---------------|------------|------|
    | Constitution | Layer 1: Overview | MASTER.md, PROJECT_RULES.md |
    | Specification | Layer 2: SoT | STATE_MACHINE.md, DATA_SCHEMA.md, etc. |
    | Plan | Layer 3: Dev-Guides | API_DEVELOPMENT_FLOW.md |
    | Tasks | Layer 3: Dev-Guides | 任务清单 |
    | Implement | Layer 4: Architecture | 代码实现 |
  </mapping>

  <rules>
    - 所有规范必须与 SoT 文档一致
    - 涉及 SoT 变更必须走 OpenSpec 流程
    - 代码生成前必须通过 SoT Guard 验证
  </rules>
</asdd_integration>

<!-- ======================================================
     4. 使用示例 (Usage Examples)
====================================================== -->
<usage>
  示例 1: 完整工作流
  「
  我要添加日报批量导出 Excel 功能。

  /speckit.constitution
  [确认宪法约束]

  /speckit.specify
  [生成规范文档]

  /speckit.plan
  [制定技术方案]

  /speckit.tasks
  [分解任务清单]

  /speckit.implement
  [执行实现]
  」

  示例 2: 从中间阶段开始
  「
  我已经有需求规范了,直接进入规划阶段。

  /speckit.plan
  requirement = "添加日报批量导出,支持按状态和日期筛选,输出 xlsx 格式"
  」

  示例 3: 单任务执行
  「
  /speckit.implement task=2
  只执行任务清单中的第 2 个任务
  」
</usage>

<!-- ======================================================
     5. 版本记录 (Version Notes)
====================================================== -->
<VERSION_NOTES>
  ### v1.0 (2025-12-17)
  - 初始版本
  - 基于 GitHub Spec-Kit 适配
  - 五阶段工作流: Constitution → Specify → Plan → Tasks → Implement
  - 与 ASDD 4 层架构集成
  - 添加辅助命令: clarify, analyze, checklist, status
</VERSION_NOTES>

</skill>

Related Skills

speckit-documentation-engineer.agent

16
from diegosouzapw/awesome-omni-skill

Expert documentation engineer specializing in technical documentation, API docs, developer guides, and documentation-as-code. Creates maintainable, searchable documentation that developers actually use.

spec_driven_development.constitution

16
from diegosouzapw/awesome-omni-skill

Creates foundational governance principles and development guidelines for the project. Use when starting a new project or establishing standards.

spec-interview

16
from diegosouzapw/awesome-omni-skill

通过系统性访谈完善技术规格文档,访谈完成后自动创建 OpenSpec proposal。适用于需求细化、技术方案设计、规范驱动开发等场景。

documentation-specialist

16
from diegosouzapw/awesome-omni-skill

文档专家。专注于技术文档编写、API 文档生成、README 优化和文档维护。提供清晰的文档结构、规范的格式和用户友好的内容。

terraform-specialist

16
from diegosouzapw/awesome-omni-skill

Expert Terraform/OpenTofu specialist mastering advanced IaC automation, state management, and enterprise infrastructure patterns.

sync-specs

16
from diegosouzapw/awesome-omni-skill

Sync ALL increment specifications to living docs structure by default. Creates FS-XXX folders for each increment. Use with increment ID to sync single increment.

specify-solution

16
from diegosouzapw/awesome-omni-skill

Create and validate solution design documents (SDD). Use when designing architecture, defining interfaces, documenting technical decisions, analyzing system components, or working on solution.md files in .start/specs/. Includes validation checklist, consistency verification, and overlap detection.

specialist-devops-infra

16
from diegosouzapw/awesome-omni-skill

Engenharia DevOps para automação, CI/CD e infraestrutura como código

section-special

16
from diegosouzapw/awesome-omni-skill

Generate comprehensive R tutorials for specialized applications (health economics, qualitative research, signal processing, environmental epidemiology) with theory + practice workflow. Use when: (1) User requests domain-specific tutorials, (2) File names match [number]-[topic].rmd pattern, (3) Keywords: TreeAge, CEA, text mining, wavelet, VMD, DLNM, WQS, BKMR.

pronunciation-specialist

16
from diegosouzapw/awesome-omni-skill

Scan lyrics for pronunciation risks, prevent Suno mispronunciations

project-specification-writer

16
from diegosouzapw/awesome-omni-skill

Generate a complete software specification document for the current project/repo, including architecture, data model, key processes, pseudocode, and Mermaid diagrams (context, container/deployment, module relations, sequence, ER, class, flowchart, state).

fda-consultant-specialist

16
from diegosouzapw/awesome-omni-skill

Senior FDA consultant and specialist for medical device companies including HIPAA compliance and requirement management. Provides FDA pathway expertise, QSR compliance, cybersecurity guidance, and regulatory submission support. Use for FDA submission planning, QSR compliance assessments, HIPAA evaluations, and FDA regulatory strategy development.