Spec Workflow MCP

English | 简体中文

Guide AI to systematically complete software development through a structured Requirements → Design → Tasks workflow, ensuring code implementation stays aligned with business needs.

Why Use It?

❌ Without Spec Workflow

• AI jumps randomly between tasks, lacking systematic approach
• Requirements disconnect from actual code implementation
• Scattered documentation, difficult to track project progress
• Missing design decision records

✅ With Spec Workflow

• AI completes tasks sequentially, maintaining focus and context
• Complete traceability from user stories to code implementation
• Standardized document templates with automatic progress management
• Each stage requires confirmation, ensuring correct direction
• Persistent progress: Continue from where you left off with check, even in new conversations

Recent Updates

v1.0.7

• 🎯 Improved reliability for most models to manage tasks with spec workflow

v1.0.6

• ✨ Batch task completion: Complete multiple tasks at once for faster progress on large projects

v1.0.5

• 🐛 Edge case fixes: Distinguish between "task not found" and "task already completed" to prevent workflow interruption

v1.0.4

• ✅ Task management: Added task completion tracking for systematic project progression

v1.0.3

• 🎉 Initial release: Core workflow framework for Requirements → Design → Tasks

Quick Start

1. Install (Claude Code Example)

claude mcp add spec-workflow-mcp -s user -- npx -y spec-workflow-mcp@latest

See full installation guide for other clients.

2. Start a New Project

"Help me use spec workflow to create a user authentication system"

3. Continue Existing Project

"Use spec workflow to check ./my-project"

The AI will automatically detect project status and continue from where it left off.

Workflow Example

1. You describe requirements

You: "I need to build a user authentication system"

2. AI creates structured documents

AI: "I'll help you create spec workflow for user authentication..."

📝 requirements.md - User stories and functional requirements
🎨 design.md - Technical architecture and design decisions
✅ tasks.md - Concrete implementation task list

3. Review and implement step by step

After each stage, the AI requests your confirmation before proceeding, ensuring the project stays on the right track.

Document Organization

Basic Structure

my-project/specs/
├── requirements.md              # Requirements: user stories, functional specs
├── design.md                    # Design: architecture, APIs, data models
├── tasks.md                     # Tasks: numbered implementation steps
└── .workflow-confirmations.json # Status: automatic progress tracking

Multi-module Projects

my-project/specs/
├── user-authentication/         # Auth module
├── payment-system/             # Payment module
└── notification-service/       # Notification module

You can specify any directory: "Use spec workflow to create auth docs in ./src/features/auth"

AI Usage Guide

🤖 Make AI Use This Tool Better

Strongly recommended to add the following prompt to your AI assistant configuration. Without it, AI may:

• ❌ Not know when to invoke Spec Workflow
• ❌ Forget to manage task progress, causing disorganized work
• ❌ Not utilize Spec Workflow for systematic documentation
• ❌ Unable to continuously track project status

With this configuration, AI will intelligently use Spec Workflow to manage the entire development process.

Configuration Note: Please modify the following based on your needs:

1. Change ./specs to your preferred documentation directory path
2. Change "English" to your preferred documentation language (e.g., "Chinese")

# Spec Workflow Usage Guidelines

## 1. Check Project Progress
When user mentions continuing previous project or is unsure about current progress, proactively use:
specs-workflow tool with action.type="check" and path="./specs"

## 2. Documentation Language
All spec workflow documents should be written in English consistently, including all content in requirements, design, and task documents.

## 3. Documentation Directory
All spec workflow documents should be placed in ./specs directory to maintain consistent project documentation organization.

## 4. Task Management
Always use the following to manage task progress:
specs-workflow tool with action.type="complete_task" and taskNumber="current task number"
Follow the workflow guidance to continue working until all tasks are completed.

## 5. Best Practices
- Proactive progress check: When user says "continue from last time", first use check to see current status
- Language consistency: Use the same language throughout all project documents
- Flexible structure: Choose single-module or multi-module organization based on project scale
- Task granularity: Each task should be completable within 1-2 hours

Installation

Requirements

• Node.js ≥ v18.0.0
• npm or yarn
• Claude Desktop or any MCP-compatible client

Install in Different MCP Clients

Claude Code (Recommended)

Use the Claude CLI to add the MCP server:

claude mcp add spec-workflow-mcp -s user -- npx -y spec-workflow-mcp@latest

Claude Desktop

Add to your Claude Desktop configuration:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%/Claude/claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "spec-workflow": {
      "command": "npx",
      "args": ["-y", "spec-workflow-mcp@latest"]
    }
  }
}

Cursor

Add to your Cursor configuration (~/.cursor/config.json):

{
  "mcpServers": {
    "spec-workflow": {
      "command": "npx",
      "args": ["-y", "spec-workflow-mcp@latest"]
    }
  }
}

Cline

Use Cline's MCP server management UI to add the server:

1. Open VS Code with Cline extension
2. Open Cline settings (gear icon)
3. Navigate to MCP Servers section
4. Add new server with:
   • Command: npx
   • Arguments: -y spec-workflow-mcp@latest

Windsurf (Codeium)

Add to your Windsurf configuration (~/.codeium/windsurf/mcp_config.json):

{
  "mcpServers": {
    "spec-workflow": {
      "command": "npx",
      "args": ["-y", "spec-workflow-mcp@latest"],
      "env": {},
      "autoApprove": [],
      "disabled": false,
      "timeout": 60,
      "transportType": "stdio"
    }
  }
}

VS Code (with MCP extension)

Add to your VS Code settings (settings.json):

{
  "mcp.servers": {
    "spec-workflow": {
      "command": "npx",
      "args": ["-y", "spec-workflow-mcp@latest"]
    }
  }
}

Zed

Add to your Zed configuration (~/.config/zed/settings.json):

{
  "assistant": {
    "version": "2",
    "mcp": {
      "servers": {
        "spec-workflow": {
          "command": "npx",
          "args": ["-y", "spec-workflow-mcp@latest"]
        }
      }
    }
  }
}

Install from Source

git clone https://github.com/kingkongshot/specs-mcp.git
cd specs-mcp
npm install
npm run build

Then add to Claude Desktop configuration:

{
  "mcpServers": {
    "spec-workflow": {
      "command": "node",
      "args": ["/absolute/path/to/specs-mcp/dist/index.js"]
    }
  }
}

Links

• GitHub Repository
• NPM Package
• Report Issues

License

MIT License