AI Expert Workflow

MCPAI代理

提供方 未标注

AI代理 stdio · 本地进程 未标注 社区 / 第三方
查看接入方式

全面解析AI Expert WorkflowMCP Server的核心功能、安装配置和实用案例。作为顶级Model Context Protocol服务器,AI Expert Workflow能让AI助手访问实时数据、执行操作,为您提供更智能的工作体验和自动化解决方案。

一句话选型结论

按资料字段整理

AI Expert Workflow 属于AI代理类 MCP 服务器,stdio · 本地进程,鉴权方式未标注,有可直接执行的安装命令。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类AI代理 传输方式stdio · 本地进程 鉴权未标注 部署方式本地部署
GitHub 星标 未标注 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

AI代理
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

安装与配置

stdio · 本地进程 未标注
客户端在本机拉起子进程,数据不出本机
npx ai-expert-workflow-mcp

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    stdio 服务在本地起进程,需要先装好 Node.js,再按安装命令拉取依赖。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
给代理接外部能力

让代理在任务中途调用搜索、数据库等真实工具。

多步任务编排

把一串操作拆成步骤,由代理按顺序执行。

自动化重复流程

把固定流程交给代理跑,减少手工操作。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在AI代理场景里让模型直接调用外部能力的人。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 完全不想在本地装环境的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
AI Expert Workflow当前 未标注 stdio · 本地进程 本地部署 有 2026-05-17
Filesystem 560c4363 13 未标注 本地部署 无 2026-05-17
Server Everything (everything) 11 stdio · 本地进程 本地部署 有 2026-05-17
Time 7365cb6d 2 SSE · 服务端推送 远程部署 有 2026-05-18
Everything Claude Code 未标注 stdio · 本地进程 本地部署 有 2026-05-17
Openmemory 557e1585 未标注 stdio · 本地进程 本地部署 有 2026-05-17

服务器档案

AI代理
字段来自项目仓库与公开索引,中文说明由本站标注
提供方未标注上游标注的维护方 未标注
分类AI代理本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式未标注远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · Node.js本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验2026-05-17上游最后核验时间

详细介绍

21.1 KB
来自项目仓库与上游目录的原文,未做删改

AI Expert Workflow MCP

An MCP (Model Context Protocol) server that implements the AI Expert Workflow using OpenRouter API for AI capabilities. This project allows you to consult with AI experts in product management, UX design, and software architecture to plan and develop your projects. It can be used standalone or with Task Master for task management (optional).

How AI Expert Workflow Interacts With You

The AI Expert Workflow is designed to provide a natural, conversational experience that guides you through the product development process:

Conversation Flow

The system engages in conversation with you until you're ready to generate a document:

  1. You Control Document Generation
    - The AI will chat with you until you explicitly request document generation with phrases like "generate document" or "create PRD"
    - If you request a document before all important topics are covered, the AI will let you know what's missing and continue the conversation
    - Only when all necessary topics are covered will the AI generate the document

  2. Comprehensive Topic Coverage
    - The AI guides you through all important aspects of your product
    - Each expert focuses on specific topics (product vision, user personas, technical architecture, etc.)
    - The AI doesn't move to the next topic until the current one is thoroughly explored

  3. Verification Before Completion
    - Before concluding each stage, the AI verifies that all your requirements are accurately captured
    - The AI asks for your confirmation before generating any document
    - You have full control over when to move forward

Document Storage

When you're ready to generate documents:

  • Each expert's document is saved in the main directory (e.g., PRD.md, UX_Design_Document.md)
  • The scripts/prd.txt file is continuously updated throughout all phases:
  • First, it contains the Product Manager's document
  • Then, the UX Designer's content is added with appropriate section headers
  • Finally, the Software Architect's content is added to complete the specification
  • This ensures that by the end of the workflow, scripts/prd.txt contains a comprehensive specification with input from all three experts
  • Individual expert contributions are also saved separately as reference files

Important: The scripts/prd.txt file is continuously improved throughout the workflow, making it ready for Task Master at any point, but it's most valuable after all three expert phases are complete.

🚀 Quick Start for End Users

Standalone Usage (Recommended)

  1. Install AI Expert Workflow globally:bash npm install -g ai-expert-workflow-mcp

  2. Get an OpenRouter API key:
    - OpenRouter API key for AI Expert Workflow: Get one here

  3. Generate a PRD for your project:```bash
    # Set your API key
    export OPENROUTER_API_KEY=your_key_here

# Generate a PRD
npx ai-expert-workflow-generate "Your detailed project description"
```

  1. Review your PRD in prd.md

Optional: Task Master Integration

If you also want to use Task Master for task management (optional):

  1. Install Task Master globally:bash npm install -g task-master-ai

  2. Get the additional API keys:
    - Anthropic API key for Task Master AI: Get one here
    - Perplexity API key (optional) for Task Master AI enhanced research capability: Get one here

  3. Configure your Cursor settings with both MCPs:json { "mcpServers": { "ai-expert-workflow": { "command": "npx", "args": ["-y", "ai-expert-workflow-mcp"], "env": { "OPENROUTER_API_KEY": "YOUR_OPENROUTER_API_KEY_HERE", "OPENROUTER_MODEL": "tngtech/deepseek-r1t-chimera:free", "MAX_TOKENS": 8000, "TEMPERATURE": 0.7 } }, "taskmaster-ai": { "command": "npx", "args": ["-y", "task-master-ai"], "env": { "ANTHROPIC_API_KEY": "YOUR_ANTHROPIC_API_KEY_HERE", "PERPLEXITY_API_KEY": "YOUR_PERPLEXITY_API_KEY_HERE", "MODEL": "claude-3-sonnet-20240229", "PERPLEXITY_MODEL": "sonar-pro", "MAX_TOKENS": 64000, "TEMPERATURE": 0.2, "DEFAULT_SUBTASKS": 5, "DEFAULT_PRIORITY": "medium" } } } }

Note on models: With OpenRouter, you can choose from many AI models beyond Claude:
- For creative tasks: openai/gpt-4o, anthropic/claude-3-opus-20240229
- For balanced performance: anthropic/claude-3-sonnet-20240229, mistral/mistral-large
- For faster, cost-effective options: openai/gpt-3.5-turbo, anthropic/claude-3-haiku-20240307
- Other options: google/gemini-pro, meta/llama-3-70b, cohere/command-r

See the complete list of OpenRouter models for all available options.

  1. Enable the MCPs in your Cursor settings.

Complete Workflow Examples

Option 1: Standalone PRD Generation (Recommended)```bash

Install AI Expert Workflow globally

npm install -g ai-expert-workflow-mcp

Set your OpenRouter API key

export OPENROUTER_API_KEY=your_openrouter_key_here

Generate a PRD with one command

npx ai-expert-workflow-generate "I want to build a recipe app that helps users find recipes based on ingredients they already have at home. Target users are home cooks who want to reduce food waste and save money."


This single command will:
1. Generate a comprehensive PRD document based on your description
2. Save it as `prd.md` in your current directory
3. Also save it in Task Master compatible format (at `scripts/prd.txt`) in case you want to use Task Master later

#### Option 2: Interactive Conversation Workflow

#### Step 1: Plan your Product with the AI Product Manager```
User: Can you start the AI Expert Workflow for my new mobile recipe app?

AI: I'll start the AI Expert Workflow to help you plan your mobile recipe app. What specific aspect would you like to begin with?

User: Let's create a PRD with the AI Product Manager. I want to build a recipe app that helps users find recipes based on ingredients they already have at home.

AI: [Asks clarifying questions about target users, key features, monetization, etc.]

User: [Provides detailed responses about the app vision]

Step 2: Generate the PRD```

User: Can you generate a complete PRD document based on our consultation?

AI: [Generates comprehensive PRD with MVP focus and lean startup validation plan]

AI: Document saved to PRD.md. You can now review it and use it for your development planning.


#### Step 3 (Optional): Use Task Master to Create Tasks
If you want to break down your PRD into development tasks, you can optionally use Task Master:```
User: Can you parse the PRD at scripts/prd.txt and generate tasks?

AI: [Task Master parses the PRD and creates a structured list of tasks with priorities]

User: What's the next task I should work on?

AI: [Task Master suggests the highest priority task with implementation details]

This streamlined workflow transforms your idea into a well-planned product with a comprehensive PRD. You can then optionally use Task Master to break it down into development tasks.

Features

  • Enhanced Workflow Structure: Structured conversation flow with topic tracking and stage transitions
  • Standalone PRD Generation: Generate comprehensive PRDs without any dependencies
  • Consult with AI experts in three key roles:
  • AI Product Manager
  • AI UX Designer
  • AI Software Architect
  • Topic-Based Conversations: Ensures comprehensive coverage of all important aspects
  • Progress Tracking: Automatically tracks completed topics and guides you through the process
  • Generate comprehensive documents:
  • Product Requirements Document (PRD) with MVP focus and lean startup approach
  • UX Design Document with prototype descriptions and user testing plans
  • Software Architecture Specification with functional specifications and technical design
  • Comprehensive Project Specification combining all three documents
  • Optional Task Master integration: Can be used with Task Master if you want task management
  • Powered by OpenRouter API for maximum flexibility across AI models:
  • Works with OpenAI models (GPT-4o, GPT-4-turbo, GPT-3.5-turbo)
  • Works with Anthropic models (Claude 3 Opus, Sonnet, Haiku)
  • Works with other models (Google Gemini, Mistral, Llama, Cohere, etc.)

Benefits of Using AI Expert Workflow MCP

  • Structured Approach: Guided workflow ensures comprehensive coverage of all important aspects
  • Complete Project Planning: Expert guidance through the entire product development process
  • Topic Tracking: Automatic tracking of completed topics prevents gaps in your planning
  • Seamless Transitions: Smooth transitions between experts with context preservation
  • Standalone Operation: Works completely on its own without requiring Task Master
  • Lean Startup Approach: Focus on MVP definition and validation plans
  • User-Centered Design: Create user experiences based on solid UX principles
  • Technical Excellence: Design scalable architectures that meet your requirements
  • Comprehensive Documentation: Generate complete project specifications that combine all phases
  • Optional Task Master Integration: Convert your PRD into development tasks if needed
  • Cursor Integration: Seamless workflow within your development environment
  • Model Flexibility: Choose from dozens of AI models through OpenRouter API
  • Use high-performance models (Claude Opus, GPT-4o) for complex planning
  • Use balanced models (Claude Sonnet, Mistral) for everyday work
  • Use fast models (Claude Haiku, GPT-3.5) for quick iterations

Installation

Global Installation (Recommended)

The easiest way to use AI Expert Workflow MCP is to install it directly from npm:```bash
npm install -g ai-expert-workflow-mcp


This makes the `ai-expert-workflow-mcp` command globally available in your terminal.

### Local Development

1. Clone the repository:```
   git clone https://github.com/bacoco/ai-expert-workflow-mcp.git
   cd ai-expert-workflow-mcp
   ```

2. Install dependencies:```
   npm install
   ```

3. Configure environment variables:
   - Copy `.env.example` to `.env`
   - Add your OpenRouter API key to the `.env` file (get one from https://openrouter.ai/keys)

4. Build the project:```
   npm run build
   ```

5. Start the server:```
   npm start
   ```

6. Run tests (optional):```
   # Run essential tests (recommended for quick verification)
   npm run test:essential

   # Run individual tests
   npm run test:mcp-only        # Test only the MCP server
   npm run test:openrouter-direct  # Test only the OpenRouter API

   # Run JavaScript tests
   npm test

   # Run TypeScript tests
   npm run test:ts

   # Or use the shell scripts
   ./tests/run-js-test.sh
   ./tests/run-ts-test.sh

   # Run comprehensive tests (builds and runs both JS and TS tests)
   ./tests/run-comprehensive-test.sh
   ```

   Test results will be saved to `tests/results/result_test.md` and `tests/results/result_test_ts.md`.

   > **IMPORTANT**: The tests will fail with the default API key. To verify your OpenRouter API key without modifying the `.env` file, use one of these scripts:
   >
   > Verify and update in one step:
   >```
   > npm run verify-and-update YOUR_API_KEY
   > ```
   >
   > Or run the steps separately:
   >```
   > # Verify your API key
   > npm run verify-openrouter YOUR_API_KEY
   >
   > # Update the MCP configuration
   > npm run update-config YOUR_API_KEY
   > ```
## Usage with Cursor AI

1. Add the MCP configuration to your editor:```json
{
  "mcpServers": {
    "ai-expert-workflow": {
      "command": "node",
      "args": ["dist/index.js"],
      "cwd": "${workspaceFolder}",
      "env": {
        "OPENROUTER_API_KEY": "${env:OPENROUTER_API_KEY}",
        "OPENROUTER_MODEL": "tngtech/deepseek-r1t-chimera:free",
        "MAX_TOKENS": 4000,
        "TEMPERATURE": 0.7,
        "MCP_TIMEOUT": "120000",
        "DEBUG": "mcp"
      }
    },
    "taskmaster-ai": {
      "command": "npx",
      "args": ["-y", "task-master-ai"],
      "env": {
        "OPENROUTER_API_KEY": "${env:OPENROUTER_API_KEY}",
        "OPENROUTER_MODEL": "openai/gpt-4-turbo",
        "MAX_TOKENS": 8000,
        "TEMPERATURE": 0.7,
        "DEFAULT_SUBTASKS": 5,
        "DEFAULT_PRIORITY": "medium"
      }
    }
  }
}
  1. Enable the MCP in your editor settings

  2. Use natural language to interact with the AI Expert Workflow:

Start the complete workflow:```
Can you start the AI Expert Workflow for my project?


**Work with specific experts:**```
Can you consult with the AI Product Manager to create a PRD for my project?

Move between stages:```
Let's move to the next stage


**Generate documents:**```
Can you generate a complete PRD document based on our consultation?

Check progress:```
What topics have we covered so far?
What topics do we still need to discuss?


4. After generating your PRD, optionally use Task Master to create tasks:```
Can you parse the PRD at scripts/prd.txt and generate tasks?

Enhanced AI Expert Workflow

The workflow has been enhanced with a structured conversation flow that guides users through a comprehensive product development process. It follows the "Méthode des Trois Sphères" (Three Spheres Method), a methodical approach that transforms ideas into functional products through three distinct phases:

The Three Spheres Method

The AI Expert Workflow implements the Three Spheres Method, where each expert builds upon the work of the previous one:

  1. Sphere 1: Product Definition & Architectural Foundation (AI Product Manager)
    - Establishes the solid foundation for the entire project
    - Defines the product vision, user personas, business requirements, and key features
    - Creates the initial PRD document

  2. Sphere 2: UX Design & Feature Expansion (AI UX Designer)
    - Uses the Product Manager's document as input
    - Develops the user experience and expands on each feature
    - Adds UI documentation, user journeys, and interaction patterns to the PRD

  3. Sphere 3: Technical Planning & Implementation Specifications (AI Software Architect)
    - Uses both the Product Manager's and UX Designer's documents as input
    - Transforms the vision into a concrete technical plan
    - Adds technical architecture, API specifications, and implementation tasks to the PRD

Each expert has access to the documents generated by previous experts, ensuring a progressive development process where each phase builds upon the previous one.

Conversation Flow and Document Generation

The system is designed to engage in conversation with the user until either:

  1. User Explicitly Requests Document Generation
    - The shouldGenerateDocument function (lines 101-118) checks for phrases like "generate document", "create PRD", etc.
    - If the user requests document generation, the system checks if all required topics have been covered (line 43)
    - If topics are missing, it informs the user and continues the conversation (lines 46-49)
    - Only when all topics are covered will it generate the document

  2. AI Determines Enough Information Has Been Gathered
    - Each expert prompt explicitly instructs to "Don't move to the next topic until the current one is sufficiently defined" (line 15)
    - The system tracks topic completion (lines 247-281)
    - The expert prompts include instructions to "verify with the user if all requirements are accurately captured" before concluding (line 32)
    - The prompts also instruct to ask the user "if they would like to generate the final PRD document" (line 32)

  3. Verification Before Moving On
    - All three expert prompts include instructions to verify with the user before concluding
    - The Product Manager prompt: "Before concluding, verify with the user if all requirements are accurately captured. When the user is satisfied, ask if they would like to generate the final PRD document..." (line 32)
    - The UX Designer prompt: "Before concluding, verify all UX decisions with the user and ensure alignment with the product vision. When the user is satisfied, ask if they would like to generate the final UX Design Document..." (line 30)
    - The Software Architect prompt: "Before concluding, verify the technical approach with the user and ensure it meets all requirements from previous stages. When the user is satisfied, ask if they would like to generate the final Software Specification document." (line 34)

  4. Document Storage
    - Each expert's document is saved in both the root directory (e.g., PRD.md) and the scripts directory (e.g., scripts/prd.txt)
    - The Product Manager's document is saved as scripts/prd.txt for Task Master compatibility
    - The UX Designer's document is saved as scripts/ux_design.txt
    - The Software Architect's document is saved as scripts/software_spec.txt
    - When all stages are complete, a comprehensive document combining all three experts' outputs is generated and saved as comprehensive_specification.md and scripts/comprehensive_specification.txt

1. Product Definition (AI Product Manager)

The Product Manager guides you through defining your product vision and requirements.

Required Topics:
- Product Vision: Overall concept and goals of the product
- User Personas: Detailed profiles of target users
- Business Requirements: Core requirements and constraints
- Feature Map: Key features with priorities
- Success Criteria: Metrics and KPIs for measuring success

Output: A comprehensive Product Requirements Document (PRD)

2. UX Design (AI UX Designer)

The UX Designer helps you create a user experience plan based on the PRD.

Required Topics:
- UI Documentation: Visual design and UI components
- Feature Specifications: Detailed feature descriptions
- User Journeys: User flows and paths
- Interaction Patterns: How users interact with features
- Data Requirements: Information architecture and data needs

Output: A detailed UX Design Document

3. Technical Architecture (AI Software Architect)

The Software Architect helps you define the technical implementation plan.

Required Topics:
- Technical Architecture: Overall system architecture
- API Specifications: API design and endpoints
- Implementation Tasks: Development tasks breakdown
- Database Schema: Data model and relationships
- Testing Strategy: Quality assurance approach

Output: A comprehensive Software Specification

The workflow automatically tracks which topics have been covered in each stage and guides you through a complete product development process. For more details, see Enhanced Workflow Documentation.

Standalone PRD Generation

The AI Expert Workflow is designed to work completely standalone:```bash

Install the AI Expert Workflow MCP

npm install -g ai-expert-workflow-mcp

Set your OpenRouter API key

export OPENROUTER_API_KEY=your_openrouter_key_here

Generate a PRD with the default model

npx ai-expert-workflow-generate "Your detailed project description"

Or specify a different model for more creative or complex projects

MODEL=openai/gpt-4o npx ai-expert-workflow-generate "Your detailed project description"
```

This script:
1. Generates a comprehensive PRD document based on your description
2. Saves it as prd.md in your current directory
3. Also saves it in Task Master compatible format (at scripts/prd.txt) in case you want to use Task Master later

Optional: Task Master Integration

> Note: Task Master integration is completely optional. You can use AI Expert Workflow on its own to generate PRDs.
>
> If you do want to use Task Master, you'll need these API keys:
> - OpenRouter API key for AI Expert Workflow (get one from OpenRouter)
> - Anthropic API key for Task Master AI (get one from Anthropic)
> - Perplexity API key (optional) for Task Master AI enhanced research capabilities (get one from Perplexity)

Using Task Master to Parse the PRD

If you want to convert your PRD into development tasks, you can use Task Master in one of two ways:

Option 1: MCP Integration (For Cursor users)
1. Add the Task Master MCP to your editor configuration:json "mcpServers": { "taskmaster-ai": { "command": "npx", "args": ["-y", "task-master-ai"], "env": { "ANTHROPIC_API_KEY": "YOUR_ANTHROPIC_API_KEY_HERE", "MODEL": "claude-3-sonnet-20240229" } } }
2. Then ask your AI assistant:Can you parse the PRD at scripts/prd.txt and generate tasks?

Option 2: Command Line Usage
1. Install Task Master:npm install -g task-master-ai
2. Parse the PRD:task-master parse-prd scripts/prd.txt

Utility Scripts

The project includes utility scripts to demonstrate Task Master integration:

  • src/generate-task-master-files.js: Creates sample PRD and Task Master integration files using the MCP implementation
  • src/generate-task-master-files-simple.js: Creates sample PRD and Task Master integration files without using MCP

These scripts are useful for testing and understanding how the integration works.

For More Information

See the docs/Create-MCP.md file for a complete guide on creating your own AI Expert Workflow MCP server from scratch.

For details on OpenRouter API integration, see docs/OPENROUTER.md.

For Task Master AI documentation, see Task Master on npm.

License

MIT

常见问题

7 问
答案由本站依据条目字段整理,不引用上游原文
AI Expert Workflow 怎么安装?

在客户端按下面的安装命令接入:npx ai-expert-workflow-mcp。装完在客户端配置里引用同一个包或端点。

AI Expert Workflow 怎么在客户端里配置?

它的传输方式未标注,在客户端 MCP 配置里按官方文档填写启动命令和参数,保存后重启客户端生效。

AI Expert Workflow 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

AI Expert Workflow 需要 API Key 吗?

资料没有标注鉴权方式,是否需要密钥取决于服务本身,接入前按官方文档确认。

AI Expert Workflow 能提供哪些能力?

资料显示能力标签包括 JavaScript、AI工作流、AI代理。具体工具清单以官方文档为准。

AI Expert Workflow 需要什么运行环境?

运行时为 Node.js;属于本地进程,需要在本机装好运行环境。

AI Expert Workflow 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-05-17。