Best use case
ai-ad-flow-orchestrator is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Teams using ai-ad-flow-orchestrator 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
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/ai-ad-flow-orchestrator/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How ai-ad-flow-orchestrator Compares
| Feature / Agent | ai-ad-flow-orchestrator | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/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-flow-orchestrator</name>
<version>1.1</version>
<domain>AI_AD_SYSTEM / 开发流程编排</domain>
<profile>Flow-Orchestrator / Multi-skill / Safe</profile>
──────────────────────────────────────────────
<!-- ======================================================
0. 核心使命(Mission)
====================================================== -->
<mission>
根据用户任务描述,自动识别适用的开发流程 (Flow),
输出完整的命令执行序列: /sot-check → /gen → /review → /doc
目标:
- 任务开始前必须进行 SoT 对齐检查
- 代码生成必须受 SoT 约束
- 生成的代码必须经过审查
- 流程结束时必须更新相关文档
SoT 基准: DEV_FLOW_SOT_v1.1 (7 大 Flow)
</mission>
<!-- ======================================================
1. 输入契约(Input Contract)
====================================================== -->
<input_contract>
最小输入:
{
task: string, // 任务描述,如 "实现充值审批功能"
}
可选输入:
{
flow_type?: "BE_DEV_FLOW" | "FE_DEV_FLOW" | "API_FIX_FLOW" | "TEST_HARDEN_FLOW" | "DOC_FREEZE_FLOW" | "REFACTOR_FLOW" | "FULL_FLOW",
target_module?: string, // 目标模块名,如 "topup", "reconciliation"
skip_steps?: string[], // 跳过的步骤,如 ["doc"]
}
命令映射 (DEV_FLOW_SOT §11):
- /dev-flow be → BE_DEV_FLOW
- /dev-flow fe → FE_DEV_FLOW
- /dev-flow fix → API_FIX_FLOW
- /dev-flow test → TEST_HARDEN_FLOW
- /dev-flow doc → DOC_FREEZE_FLOW
- /dev-flow full → FULL_FLOW
- /dev-flow refactor → REFACTOR_FLOW
若 task 缺失 → 输出 <halt>Missing: task description</halt> 并停止。
</input_contract>
<!-- ======================================================
2. 输出契约(Output Contract)
====================================================== -->
<output_contract>
成功输出:
{
flow_type: string, // 识别的流程类型
flow_name: string, // 流程中文名
command: string, // 对应的 /dev-flow 命令
command_sequence: [ // 命令执行序列
{ step: number, command: string, purpose: string, expected_output: string }
],
local_test_commands: string[], // 本地测试命令
freeze_checklist: string[], // 冻结检查项
}
失败输出:
{
error: string,
suggestion: string,
}
</output_contract>
<!-- ======================================================
3. Flow 识别规则 (基于 DEV_FLOW_SOT §3)
====================================================== -->
<flow_detection>
基于任务描述自动识别 Flow 类型 (7 大 Flow):
<!-- BE_DEV_FLOW 触发关键词 -->
<flow id="BE_DEV_FLOW" command="/dev-flow be">
<triggers>
- 新增后端
- 新增 API
- 新增接口
- 实现 service
- 实现服务
- 后端功能
- 新增模块
- 状态机流转
- 账本分录
- 后端开发
</triggers>
<complexity>高</complexity>
</flow>
<!-- FE_DEV_FLOW 触发关键词 -->
<flow id="FE_DEV_FLOW" command="/dev-flow fe">
<triggers>
- 新增页面
- 新增组件
- 前端功能
- UI 开发
- React 组件
- 表单页面
- 列表页面
- 前端开发
</triggers>
<complexity>中</complexity>
</flow>
<!-- API_FIX_FLOW 触发关键词 -->
<flow id="API_FIX_FLOW" command="/dev-flow fix">
<triggers>
- 修复接口
- 修复 API
- Bug 修复
- 接口问题
- 返回值错误
- 参数校验
- 接口报错
- fix bug
</triggers>
<complexity>低</complexity>
</flow>
<!-- TEST_HARDEN_FLOW 触发关键词 -->
<flow id="TEST_HARDEN_FLOW" command="/dev-flow test">
<triggers>
- 补充测试
- 测试覆盖
- 回归测试
- 边界测试
- 测试加固
- 单元测试
- 测试补齐
</triggers>
<complexity>中</complexity>
</flow>
<!-- DOC_FREEZE_FLOW 触发关键词 -->
<flow id="DOC_FREEZE_FLOW" command="/dev-flow doc">
<triggers>
- 文档审计
- 文档冻结
- 文档治理
- 更新文档
- SoT 更新
- 文档检查
</triggers>
<complexity>低</complexity>
</flow>
<!-- REFACTOR_FLOW 触发关键词 -->
<flow id="REFACTOR_FLOW" command="/dev-flow refactor">
<triggers>
- 代码重构
- 重构代码
- 优化代码
- 代码优化
- 去重复
- 提取公共
- 性能优化
- 技术债
- refactor
</triggers>
<complexity>中</complexity>
<constraints>
- 不得改变业务行为
- 不得修改 SoT 定义
- 输入/输出契约保持不变
</constraints>
</flow>
<!-- FULL_FLOW 触发关键词 -->
<flow id="FULL_FLOW" command="/dev-flow full">
<triggers>
- 完整功能
- 端到端
- 全栈开发
- 前后端
- MVP
- 新模块开发
- 完整模块
</triggers>
<complexity>高</complexity>
<composition>BE_DEV_FLOW → FE_DEV_FLOW → TEST_HARDEN_FLOW → DOC_FREEZE_FLOW</composition>
</flow>
若无法识别 → 默认 BE_DEV_FLOW 并提示用户确认。
</flow_detection>
<!-- ======================================================
4. 命令序列模板 (基于 DEV_FLOW_SOT §4-10)
====================================================== -->
<command_templates>
<!-- BE_DEV_FLOW 命令序列 (DEV_FLOW_SOT §4) -->
<template flow="BE_DEV_FLOW">
<step n="1">
<command>/sot-check docs/sot/</command>
<purpose>SoT 对齐 - 确认相关规则</purpose>
<expected>SoT 规则清单 + 依赖关系</expected>
</step>
<step n="2">
<command>/gen be "生成 {module} 的 Pydantic Schema"</command>
<purpose>Schema 层生成</purpose>
<expected>backend/schemas/{module}.py</expected>
</step>
<step n="3">
<command>/gen be "实现 {module} 的 Service 层"</command>
<purpose>Service 层生成</purpose>
<expected>backend/services/{module}_service.py</expected>
</step>
<step n="4">
<command>/gen be "实现 {module} 的 Router 层"</command>
<purpose>Router 层生成</purpose>
<expected>backend/routers/{module}.py</expected>
</step>
<step n="5">
<command>/gen test "为 {module} 生成状态机测试 + API 测试"</command>
<purpose>测试生成</purpose>
<expected>backend/tests/services/test_{module}_service.py, backend/tests/api/test_{module}_api.py</expected>
</step>
<step n="6">
<command>/review backend/services/{module}_service.py</command>
<purpose>代码审查 (Service 层)</purpose>
<expected>审查报告 + 修复建议</expected>
</step>
<local_test>
pytest backend/tests/services/test_{module}_service.py -v
pytest backend/tests/api/test_{module}_api.py -v
mypy backend/services/{module}_service.py
</local_test>
</template>
<!-- FE_DEV_FLOW 命令序列 (DEV_FLOW_SOT §5) -->
<template flow="FE_DEV_FLOW">
<step n="1">
<command>/sot-check docs/sot/API_SOT.md</command>
<purpose>API 契约确认</purpose>
<expected>相关 API 端点列表 + 数据结构</expected>
</step>
<step n="2">
<command>/gen fe "生成 {module} 的 API Client"</command>
<purpose>API Client 生成</purpose>
<expected>frontend/src/lib/api/{module}.ts</expected>
</step>
<step n="3">
<command>/gen fe "创建 {component} 页面/组件"</command>
<purpose>组件/页面生成</purpose>
<expected>frontend/src/modules/{module}/</expected>
</step>
<step n="4">
<command>/gen test "为 {component} 生成前端测试"</command>
<purpose>测试生成</purpose>
<expected>frontend/tests/{module}/</expected>
</step>
<step n="5">
<command>/review frontend/src/modules/{module}/</command>
<purpose>组件审查</purpose>
<expected>审查报告</expected>
</step>
<local_test>
cd frontend && npm run type-check
npm run lint
npm run test -- --testPathPattern={module}
</local_test>
</template>
<!-- API_FIX_FLOW 命令序列 (DEV_FLOW_SOT §6) -->
<template flow="API_FIX_FLOW">
<step n="1">
<command>/sot-check {target_file}</command>
<purpose>问题定位 - 对比 API_SOT.md 识别偏差点</purpose>
<expected>问题清单 + SoT 规则引用</expected>
</step>
<step n="2">
<command>/gen be "修复 {issue_description}"</command>
<purpose>修复生成</purpose>
<expected>修改后的 Router/Service 代码</expected>
</step>
<step n="3">
<command>/gen test "为 {target_file} 生成回归测试"</command>
<purpose>回归测试</purpose>
<expected>回归测试用例</expected>
</step>
<step n="4">
<command>/review {target_file}</command>
<purpose>验证审查</purpose>
<expected>审查报告</expected>
</step>
<local_test>
pytest backend/tests/api/test_{module}_api.py::{test_case} -v
</local_test>
</template>
<!-- TEST_HARDEN_FLOW 命令序列 (DEV_FLOW_SOT §7) -->
<template flow="TEST_HARDEN_FLOW">
<step n="1">
<command>/sot-check backend/tests/</command>
<purpose>覆盖分析</purpose>
<expected>测试缺口清单</expected>
</step>
<step n="2">
<command>/gen test "补齐 {entity} 状态机测试"</command>
<purpose>状态机测试</purpose>
<expected>状态机测试用例</expected>
</step>
<step n="3">
<command>/gen test "补齐 {module} 边界条件测试"</command>
<purpose>边界测试</purpose>
<expected>边界测试用例</expected>
</step>
<step n="4">
<command>/review backend/tests/</command>
<purpose>覆盖验证</purpose>
<expected>测试审查报告</expected>
</step>
<local_test>
pytest backend/tests/ -v
pytest backend/tests/ --cov=backend --cov-report=html
</local_test>
</template>
<!-- DOC_FREEZE_FLOW 命令序列 (DEV_FLOW_SOT §8) -->
<template flow="DOC_FREEZE_FLOW">
<step n="1">
<command>/doc docs/</command>
<purpose>文档审计</purpose>
<expected>P0/P1/P2 问题清单</expected>
</step>
<step n="2">
<command>/sot-check docs/sot/</command>
<purpose>SoT 一致性检查</purpose>
<expected>SoT 一致性报告</expected>
</step>
<step n="3">
<command>/doc --freeze {module}</command>
<purpose>Freeze 报告生成</purpose>
<expected>docs/reports/{module}_FREEZE_REPORT.md</expected>
</step>
<local_test>
# 无代码测试,人工确认文档准确性
</local_test>
</template>
<!-- REFACTOR_FLOW 命令序列 (DEV_FLOW_SOT §9) -->
<template flow="REFACTOR_FLOW">
<constraints>
⚠️ 不得改变业务行为
⚠️ 不得修改 SoT 定义
⚠️ 输入/输出契约保持不变
</constraints>
<entry_conditions>
- 测试覆盖率 >= 80%
- CI 通过
- SoT 文档 frozen
- 无 P0 级 Bug
</entry_conditions>
<step n="1">
<command>pytest backend/tests/ --tb=short > refactor_baseline.txt</command>
<purpose>快照基线</purpose>
<expected>测试快照 + API 响应样本</expected>
</step>
<step n="2">
<command>/sot-check {target_files}</command>
<purpose>SoT 对齐检查</purpose>
<expected>SoT 合规报告 + 技术债清单</expected>
</step>
<step n="3">
<command>/review {target_files}</command>
<purpose>代码分析</purpose>
<expected>重构建议清单</expected>
</step>
<step n="4">
<command>/gen be "重构 {module} 的 {重构目标}"</command>
<purpose>重构实施</purpose>
<expected>重构后的代码</expected>
</step>
<step n="5">
<command>/sot-check {target_files}</command>
<purpose>等价验证</purpose>
<expected>SoT 合规 + 测试通过</expected>
</step>
<local_test>
pytest backend/tests/ -v > refactor_after.txt
diff refactor_baseline.txt refactor_after.txt
</local_test>
<exit_conditions>
- 所有现有测试通过
- 行为等价 (diff 无差异)
- /sot-check 无新增 P0/P1
</exit_conditions>
</template>
<!-- FULL_FLOW 命令序列 (DEV_FLOW_SOT §10) -->
<template flow="FULL_FLOW">
<composition>
Phase 1: BE_DEV_FLOW (完整 6 步)
Phase 2: FE_DEV_FLOW (完整 5 步)
Phase 3: TEST_HARDEN_FLOW (完整 4 步)
Phase 4: DOC_FREEZE_FLOW (完整 3 步)
</composition>
<step n="1" phase="backend">
<command>执行 BE_DEV_FLOW 完整流程</command>
<purpose>后端开发</purpose>
<expected>Schema + Service + Router + 后端测试</expected>
</step>
<step n="2" phase="frontend">
<command>执行 FE_DEV_FLOW 完整流程</command>
<purpose>前端开发</purpose>
<expected>API Client + 页面/组件 + 前端测试</expected>
</step>
<step n="3" phase="test">
<command>执行 TEST_HARDEN_FLOW 完整流程</command>
<purpose>测试加固</purpose>
<expected>状态机测试 + 边界测试 + 覆盖率报告</expected>
</step>
<step n="4" phase="doc">
<command>执行 DOC_FREEZE_FLOW 完整流程</command>
<purpose>文档冻结</purpose>
<expected>文档审计报告 + Freeze 报告</expected>
</step>
<local_test>
# 后端测试
pytest backend/tests/ -v
# 前端测试
cd frontend && npm run test
# 覆盖率
pytest backend/tests/ --cov=backend --cov-report=html
</local_test>
</template>
</command_templates>
<!-- ======================================================
5. 执行流程
====================================================== -->
<execution_workflow>
Phase 1: 任务分析
─────────────────
1. 接收用户任务描述
2. 提取关键词,匹配 Flow 类型 (7 大 Flow)
3. 若用户指定 flow_type,使用指定值
4. 输出 Flow 类型确认 + 对应 /dev-flow 命令
Phase 2: 命令序列生成
─────────────────────
1. 加载对应 Flow 模板 (参考 DEV_FLOW_SOT)
2. 替换 {module}, {component} 等占位符
3. 若用户指定 skip_steps,移除对应步骤
4. 输出完整命令序列
Phase 3: 执行指导
─────────────────
1. 输出命令执行顺序
2. 提示每步的预期产出
3. 提供本地测试命令
4. 输出冻结检查清单
</execution_workflow>
<!-- ======================================================
6. 输出格式
====================================================== -->
<output_format>
## Flow Orchestration Report
### 任务识别
- **任务**: {task}
- **Flow 类型**: {flow_type}
- **Flow 名称**: {flow_name}
- **命令**: {command}
- **复杂度**: {complexity}
- **SoT 基准**: DEV_FLOW_SOT v1.1
### 命令执行序列
| Step | 命令 | 目的 | 预期产出 |
|------|------|------|----------|
| 1 | ... | ... | ... |
### 执行提示
```
⚠️ 命令不支持管道串联,请逐条执行:
Step 1: /sot-check ...
→ 确认 SoT 规则后继续
Step 2: /gen be ...
→ 检查生成代码后继续
...
```
### 本地测试
```bash
# 代码测试
pytest backend/tests/ -v
# 类型检查
mypy backend/services/{module}_service.py
```
### 冻结检查清单
- [ ] 所有 SoT 规则已遵循
- [ ] 代码审查通过
- [ ] 本地测试通过
- [ ] 相关文档已更新
</output_format>
<!-- ======================================================
7. 错误处理
====================================================== -->
<error_handling>
<error code="FLOW_UNKNOWN">
<condition>无法识别 Flow 类型</condition>
<action>默认 BE_DEV_FLOW,提示用户确认</action>
</error>
<error code="MODULE_MISSING">
<condition>任务未指明目标模块</condition>
<action>询问用户补充模块名</action>
</error>
<error code="SOT_VIOLATION">
<condition>/sot-check 发现 P0 违规</condition>
<action>中断流程,输出违规详情,等待修复</action>
</error>
<error code="REFACTOR_BEHAVIOR_CHANGE">
<condition>REFACTOR_FLOW 检测到业务行为变化</condition>
<action>中断流程,提示用户确认是否改为 BE_DEV_FLOW</action>
</error>
</error_handling>
<!-- ======================================================
8. 调用示例
====================================================== -->
<examples>
<example name="后端新功能">
<input>
task: "实现充值审批功能"
</input>
<output>
flow_type: BE_DEV_FLOW
command: /dev-flow be 实现充值审批功能
command_sequence:
1. /sot-check docs/sot/
2. /gen be "生成 topup_approval 的 Pydantic Schema"
3. /gen be "实现 topup_approval 的 Service 层"
4. /gen be "实现 topup_approval 的 Router 层"
5. /gen test "为 topup_approval 生成状态机测试 + API 测试"
6. /review backend/services/topup_approval_service.py
</output>
</example>
<example name="接口修复">
<input>
task: "修复日报导出接口返回空数据问题"
</input>
<output>
flow_type: API_FIX_FLOW
command: /dev-flow fix 修复日报导出接口返回空数据问题
command_sequence:
1. /sot-check backend/routers/daily_report.py
2. /gen be "修复日报导出接口返回空数据问题"
3. /gen test "为日报导出接口生成回归测试"
4. /review backend/routers/daily_report.py
</output>
</example>
<example name="代码重构">
<input>
task: "重构 topup_service.py 的审批逻辑"
</input>
<output>
flow_type: REFACTOR_FLOW
command: /dev-flow refactor 重构 topup_service.py 的审批逻辑
constraints:
- 不得改变业务行为
- 不得修改 SoT 定义
command_sequence:
1. pytest backend/tests/ --tb=short > refactor_baseline.txt
2. /sot-check backend/services/topup_service.py
3. /review backend/services/topup_service.py
4. /gen be "重构 topup_service.py 的审批逻辑"
5. /sot-check backend/services/topup_service.py
6. pytest + diff 验证
</output>
</example>
<example name="完整功能开发">
<input>
task: "实现对账模块"
</input>
<output>
flow_type: FULL_FLOW
command: /dev-flow full 实现对账模块
phases:
Phase 1: BE_DEV_FLOW (后端)
Phase 2: FE_DEV_FLOW (前端)
Phase 3: TEST_HARDEN_FLOW (测试加固)
Phase 4: DOC_FREEZE_FLOW (文档冻结)
</output>
</example>
</examples>
</skill>Related Skills
analyzing-taint-flow
Tracks untrusted input propagation from sources to sinks in binary code to identify injection vulnerabilities. Use when analyzing data flow, tracing user input to dangerous functions, or detecting command/SQL injection.
ai-native-builder-workflow
A complete end-to-end framework for non-technical product managers to build and ship software using AI coding agents. Use this when starting a side project, building a prototype, or automating internal tools without an engineering team.
ai-code-factory-workflow
No description provided.
ai-ad-doc-orchestrator
No description provided.
agile-workflow
Orchestrate agile development workflows by invoking commands in sequence with checkpoint-based flow control. This skill should be used when the user asks to 'run the workflow', 'continue working', 'what's next', 'complete the task cycle', 'start my day', 'end the sprint', 'implement the next task', or wants guided step-by-step development assistance. Keywords: workflow, orchestrate, agile, task cycle, sprint, daily, implement, review, PR, standup, retrospective.
agency-workflow-patterns
Master orchestration patterns, multi-agent coordination, and effective workflow composition using the Agency plugin's 51+ specialized agents. Activate when planning complex implementations, coordinating multiple agents, or optimizing development workflows.
AEM Development Workflow
Required workflow for AEM EDS development - linting, testing, preview validation, and PRs with demo links. Use when developing blocks or features for AEM Edge Delivery Services projects, before doing anything else.
advanced-workflows
Multi-tool orchestration patterns for complex Bluera Knowledge operations. Teaches progressive library exploration, adding libraries with job monitoring, handling large result sets, multi-store searches, and error recovery workflows.
workflow-orchestration-patterns
Design durable workflows with Temporal for distributed systems. Covers workflow vs activity separation, saga patterns, state management, and determinism constraints. Use when building long-running ...
workflow-creator
Meta-skill for designing and creating Antigravity workflows. Interviews user, proposes optimal structure, checks for duplicates, and ensures workflows integrate with existing skills.
map-workflows-guide
Comprehensive guide for choosing the right MAP workflow based on task type and requirements
android-ux-flows
This skill is used to design Android user flows and screen structures that match the existing app patterns and keep forms, lists and navigation clear.