SDS (Software Design Specification) Generator

A tool that converts natural language project descriptions into comprehensive software design specifications. Now featuring a modular architecture for improved maintainability and extensibility.

Features

• Natural Language Input: Supports both Korean and English project descriptions
• Auto Project Type Detection: Automatically detects Mobile, Web, Backend projects
• Smart API Selection: Intelligently selects best AI API based on task type and performance
• Tech Stack Selection: Multiple technology stack options for each platform
• Language-Specific Code Generation: Generates correct syntax for Swift, Dart, Java, Python, JavaScript
• Dynamic File Extensions: Creates .swift, .dart, .java, .py files based on selected tech stack
• Complete Specification: Generates detailed specifications with functions and test cases
• Development Files: Auto-generates structured files in .sds directory
• MCP Server: Compatible with Claude Code and other MCP clients
• Modular Architecture: Clean separation of concerns for better maintainability

Installation

Global Installation (Recommended)

npm install -g sds-generator

Local Installation

npm install sds-generator

Setup API Keys

cp .env.example .env
# Edit .env file and add your API keys

Usage

Global Installation

sds "project description"

Local Installation

npx sds-generator "project description"
# or
node sds.js "project description"

Example

sds "I want to create a mobile blood sugar monitoring app. Users should be able to input and record blood sugar levels, view trends in graphs, set target ranges, and receive notifications."

Process:

1. Auto-detect project type (mobile)
2. Select tech stack (React Native, Flutter, Swift, Kotlin)
3. Generate and output specification
4. Create development files in .sds/ directory

Generated Files

Markdown Specification

• specification.md: Display-ready specification with tables

.sds Directory (Development)

• development.json: Tech stack and implementation details
• tasks.json: TaskMaster AI compatible task list
• api.json: OpenAPI specification
• database.json: Database schema
• README.md: Development guide
• Language-specific module files: .swift, .dart, .java, .py, .js based on selected tech stack

Supported Project Types

Mobile

• React Native
• Flutter
• Native iOS (Swift)
• Native Android (Kotlin)

Web

• React/Next.js
• Vue/Nuxt

Backend

• Node.js/Express
• Python/FastAPI

Example Output

🎯 Detected project type: mobile

🔧 Select tech stack for mobile project:
1. React Native
2. Flutter
3. Native iOS (Swift)
4. Native Android (Kotlin)

Select tech stack (1-4): 2

✅ Selected tech stack: Flutter

Generated specification includes:

• Tech stack information
• Dependencies list
• Module function table (Module | Function | Design Spec | Function Definition | Remarks)
• Detailed module descriptions

Architecture

Modular Structure

• lib/errors.js: Custom error classes and centralized error handling
• lib/config.js: Configuration management and logging system
• lib/api-client.js: Unified API client with smart selection
• mcp-server.js: MCP server implementation
• sds.js: CLI interface

Supported APIs

• Anthropic Claude: Set ANTHROPIC_API_KEY (Recommended)
• OpenAI GPT: Set OPENAI_API_KEY
• Perplexity: Set PERPLEXITY_API_KEY

MCP Integration

Use as MCP server with Claude Code:

sds --mcp

License

MIT