Skip to main content
Glama

System Designer MCP Server

by chevyfsa
GitHub
TypeScript
MIT License
1
  • Apple
OverviewInspectNewEndpointsSchemaRelated ServersReviewsScore
Need Help?View Source CodeReport Issue

MseeP.ai Security Assessment Badge

System Designer MCP Server

CodeRabbit Pull Request Reviews Verified on MseeP

A Model Context Protocol (MCP) server that provides AI agents with tools to create, validate, and export UML system models and System Runtime bundles. Built with a tool-based approach that empowers LLMs to generate complete, executable System Runtime applications.

📚 Documentation

Features

Core MSON Tools

  • create_mson_model: Create and validate MSON models from structured data

  • validate_mson_model: Validate MSON model consistency and completeness

  • generate_uml_diagram: Generate UML diagrams in PlantUML and Mermaid formats

  • export_to_system_designer: Export models to System Designer application format

System Runtime Tools

  • create_system_runtime_bundle: Convert MSON models to complete System Runtime bundles

  • validate_system_runtime_bundle: Validate System Runtime bundles for correctness and compatibility

Key Capabilities

  • ✅ Tool-Based Architecture: LLMs handle understanding, server handles validation/formatting

  • ✅ Type Safety: Comprehensive Zod schema validation for all inputs and outputs

  • ✅ System Runtime Integration: Full support for System Runtime bundle generation and validation

  • ✅ Bidirectional Relationships: Automatic bidirectional relationship creation

  • ✅ Multiple Inheritance: Support for classes implementing multiple interfaces

  • ✅ Multiple UML Formats: Support for both PlantUML and Mermaid diagram generation

  • ✅ System Designer Integration: Direct export to System Designer macOS application

  • ✅ Comprehensive Testing: 46 tests with 302 expect() calls covering all functionality

  • ✅ Critical Bug Fixes: Resolved SDMCP-001 through SDMCP-005 (property preservation, ID management, validation consistency)

  • ✅ Streamable HTTP Transport: Modern MCP protocol (2025-03-26 specification)

  • ✅ Stateless Operation: Each request creates new server instance

  • ✅ Single Endpoint: Single /mcp endpoint handles all operations

  • ✅ Workers Optimized: Stateless design perfect for Cloudflare Workers

Installation

Prerequisites

  • Bun JavaScript runtime

  • Node.js compatibility through Bun

Setup

# Clone the repository git clone https://github.com/chevyphillip/system-designer-mcp.git cd system-designer-mcp # Install dependencies bun install # Build the project bun run build # Run tests bun test

Quick Start

# Clone the repository git clone https://github.com/chevyphillip/system-designer-mcp.git cd system-designer-mcp # Install dependencies bun install # Build the project bun run build # Run tests bun test

Deployment Options

The System Designer MCP Server can run in two modes:

Local Mode (stdio transport)

Start the local server:

bun run dev

Remote Mode (Cloudflare Workers)

Deploy as a remote MCP server accessible over HTTP with SSE transport:

# Test locally with Wrangler bun run dev:worker # Deploy to Cloudflare Workers bunx wrangler login bun run deploy

Your MCP server will be available at:

https://system-designer-mcp.<your-subdomain>.workers.dev

Note: Replace <your-subdomain> with your actual Cloudflare Workers subdomain.

Key Features:

  • ✅ SSE (Server-Sent Events) transport for remote access

  • ✅ No authentication required (configurable)

  • ✅ All 6 MCP tools available

  • ✅ Returns JSON data directly (no file system)

  • ✅ Automatic session management

  • ✅ CORS support for web clients

See

Usage Examples

Example tool usage:

// Create a MSON model const model = await mcpClient.callTool('create_mson_model', { name: 'Student Management System', type: 'class', description: 'A system for managing students and courses', entities: [ { id: 'student', name: 'Student', type: 'class', attributes: [ { name: 'id', type: 'string', visibility: 'private' }, { name: 'name', type: 'string', visibility: 'public' }, ], methods: [{ name: 'enroll', returnType: 'void', visibility: 'public' }], }, ], relationships: [], }); // Generate UML diagram const diagram = await mcpClient.callTool('generate_uml_diagram', { model: model.content[1].json, format: 'plantuml', }); // Export to System Designer const exported = await mcpClient.callTool('export_to_system_designer', { model: model.content[1].json, filePath: './student_system.json', }); // Create System Runtime bundle const bundle = await mcpClient.callTool('create_system_runtime_bundle', { model: model.content[1].json, version: '1.0.0', }); // Validate System Runtime bundle const validation = await mcpClient.callTool('validate_system_runtime_bundle', { bundle: bundle.content[2].text, // JSON bundle from previous step });

CLI Usage

The server includes a CLI tool for testing and model management:

# Test System Designer integration bun run src/cli.ts test-integration # Export a test model bun run src/cli.ts export-model MyModel "Test model description" # Show configuration bun run src/cli.ts config

See the CLI Guide for detailed usage instructions.

Example MSON Model Structure

{ "id": "student_system", "name": "Student Management System", "type": "class", "description": "A system for managing students and courses", "entities": [ { "id": "student", "name": "Student", "type": "class", "attributes": [ { "name": "id", "type": "string", "visibility": "private" }, { "name": "name", "type": "string", "visibility": "public" } ], "methods": [ { "name": "enroll", "parameters": [ { "name": "course", "type": "Course" } ], "returnType": "void", "visibility": "public" } ] } ], "relationships": [ { "id": "enrollment", "from": "student", "to": "course", "type": "association", "multiplicity": { "from": "1", "to": "0..*" }, "name": "enrolls in" } ] }

Tool Reference

For detailed API documentation, see the API Reference.

Available Tools

  • create_mson_model - Create and validate MSON models from structured data with automatic ID generation and relationship mapping

  • validate_mson_model - Validate MSON model consistency and completeness with detailed error messages and relationship validation

  • generate_uml_diagram - Generate UML diagrams in PlantUML and Mermaid formats

  • export_to_system_designer - Export models to System Designer application format

  • create_system_runtime_bundle - Convert MSON models to complete System Runtime bundles with schemas, models, types, behaviors, and components

  • validate_system_runtime_bundle - Validate System Runtime bundles for correctness and compatibility

Platform Integration

The server can be integrated with various platforms:

  • Claude Desktop - Native MCP integration

  • VS Code - Extension development support

  • Web Applications - React/Node.js integration

  • CLI Tools - Command-line interface

See the Integration Guide for detailed setup instructions.

Architecture

Modular Structure

The codebase follows SOLID principles with clear separation of concerns:

  • src/types.ts - TypeScript type definitions for MSON models

  • src/schemas.ts - Zod validation schemas for all data structures

  • src/tools.ts - MCP tool registration using modern SDK patterns

  • src/index.ts - Local MCP server with stdio transport (Node.js/Bun)

  • src/worker.ts - Remote MCP server with SSE transport (Cloudflare Workers)

  • src/cli.ts - Command-line interface for testing and integration

  • src/integration/ - System Designer app integration

  • src/transformers/ - MSON to System Runtime transformation logic

  • src/validators/ - System Runtime bundle validation logic

Modern MCP SDK Patterns

The server uses the modern MCP TypeScript SDK (v1.18.2) patterns:

  1. server.registerTool() - Modern tool registration API (not legacy server.tool())

  2. Zod Input Schemas - Type-safe input validation with Zod schema shapes

  3. Title Metadata - Each tool includes a title field for better UX

  4. Type Inference - Handler methods use Zod-inferred types for parameters

Tool-Based Approach

This server uses a tool-based architecture that:

  1. Empowers LLMs: The LLM handles understanding requirements and creating structured data

  2. Validates Input: Server validates structured input using comprehensive Zod schemas

  3. Processes Efficiently: Simple, fast processing without complex parsing logic

  4. Exports Flexibly: Multiple output formats for different use cases

Benefits Over Parser-Based Approaches

  • Simplicity: No complex NLP parsing to maintain

  • Flexibility: Works with any domain the LLM understands

  • Reliability: Fewer moving parts, less error-prone

  • Performance: Faster validation and processing

  • Extensibility: Easy to add new tools and features

Development

Running Tests

# Run all tests bun test # Run tests in watch mode bun test --watch

Building

# Build for production bun run build # Start production server bun start

Code Structure

src/ ├── types.ts # TypeScript type definitions for MSON models ├── schemas.ts # Zod validation schemas for all data structures ├── tools.ts # MCP tool registration using modern SDK patterns ├── index.ts # Local MCP server (stdio transport) ├── worker.ts # Remote MCP server (SSE transport for Workers) ├── cli.ts # Command-line interface ├── integration/ │ └── system-designer.ts # System Designer app integration ├── transformers/ │ └── system-runtime.ts # MSON to System Runtime transformation └── validators/ └── system-runtime.ts # System Runtime bundle validation test/ └── tool-based.test.ts # Comprehensive test suite docs/ ├── API-REFERENCE.md # Detailed API documentation ├── CLI-GUIDE.md # CLI usage guide ├── INTEGRATION-GUIDE.md # Platform integration guide └── SYSTEM-RUNTIME-INTEGRATION-GUIDE.md # System Runtime guide examples/ ├── banking-system.json # Sample banking system model ├── banking-system-plantuml.puml # PlantUML output example ├── banking-system-mermaid.md # Mermaid output example └── README.md # Example documentation CLOUDFLARE_DEPLOYMENT.md # Cloudflare Workers deployment guide test-worker.sh # Automated testing script for Workers wrangler.toml # Cloudflare Workers configuration

Integration with System Designer

The server exports models in a format compatible with the System Designer macOS application:

  1. File Export: Models are saved as JSON files

  2. Automatic Integration: Files can be imported directly into System Designer

  3. Format Compatibility: Uses MSON (Metamodel JavaScript Object Notation) format

Recent Changes & Improvements

Latest Updates (v1.0.0)

  • 🎯 Critical Bug Fixes (SDMCP-001-005): Resolved property preservation, ID mapping, validation consistency, and bundle compatibility issues

  • 🔧 Enhanced Error Messages: Added detailed, actionable error messages with specific fix suggestions and examples

  • 🛡️ Relationship Validation: Proactive validation prevents orphaned references with clear guidance

  • 🔄 Flexible Input Handling: Support for both 'properties' and 'attributes' in entity definitions

  • 📋 Comprehensive Bug Reports: Detailed documentation of all issues and resolutions in docs/BUG_REPORTS.md

  • 🆕 System Runtime Tools: Added complete System Runtime bundle creation and validation functionality

Previous Features

  • 🏗️ Modular Architecture: Complete refactoring for SOLID principles compliance

  • 🧪 Enhanced Testing: 46 tests with 302 assertions covering all functionality including edge cases

  • 🚀 Cloudflare Workers Support: Remote MCP server deployment with modern JSON-RPC transport

  • 🔄 Modern MCP SDK: Updated to latest MCP TypeScript SDK (v1.18.2) patterns

  • 📊 Documentation: Complete API reference, integration guides, and deployment documentation

Contributing

We welcome contributions! Please see our Contributing Guide for details on:

  • Setting up the development environment

  • Running tests and code quality checks

  • Code style guidelines

  • Submitting pull requests

  • Reporting issues

License

MIT License - see LICENSE file for details.

Acknowledgments

This server cannot be installed

-
security - not tested
A
license - permissive license
-
quality - not tested

How are these scores calculated?

hybrid server

The server is able to function both locally and remotely, depending on the configuration or use case.

Enables AI agents to create, validate, and export UML system models using structured MSON format. Supports generating PlantUML and Mermaid diagrams with direct export to System Designer macOS application.

  1. 📚 Documentation
    1. Features
      1. Core MSON Tools
      2. System Runtime Tools
      3. Key Capabilities
    2. Installation
      1. Prerequisites
      2. Setup
    3. Quick Start
      1. Deployment Options
        1. Local Mode (stdio transport)
        2. Remote Mode (Cloudflare Workers)
      2. Usage Examples
        1. CLI Usage
        2. Example MSON Model Structure
      3. Tool Reference
        1. Available Tools
      4. Platform Integration
        1. Architecture
          1. Modular Structure
          2. Modern MCP SDK Patterns
          3. Tool-Based Approach
          4. Benefits Over Parser-Based Approaches
        2. Development
          1. Running Tests
          2. Building
          3. Code Structure
        3. Integration with System Designer
          1. Recent Changes & Improvements
            1. Latest Updates (v1.0.0)
            2. Previous Features
          2. Contributing
            1. License
              1. Acknowledgments

                Appeared in Searches

                New MCP Servers

                MCP directory API

                We provide all the information about MCP servers via our MCP API.

                curl -X GET 'https://glama.ai/api/mcp/v1/servers/chevyfsa/system-designer-mcp'

                If you have feedback or need assistance with the MCP directory API, please join our Discord server