The Universal Model Context Protocol Platform
Transform your development workflow with a comprehensive MCP hub that unifies AI agents, enterprise tools, and custom integrations. Build, deploy, and manage MCP servers with ease.
π NEW: Custom Server Builder - Create personalized MCP servers by combining tools from multiple sources with dual STDIO/HTTP support!
- Custom Server Builder: Create personalized MCP servers in minutes
- Dual Protocol Support: Both HTTP and WebSocket/STDIO connections
- Enterprise Ready: Production-grade deployment and scaling
- Plugin Architecture: Extensible with community and custom servers
- AI-Powered: Intelligent tool discovery and workflow automation
The platform includes 9+ pre-built specialized servers with 238+ tools covering:
Category | Servers | Tools | Examples |
---|---|---|---|
Enterprise CRM | 5 servers | 151 tools | Salesforce APIs, Business Operations, DevOps |
Design Systems | 1 server | 4 tools | IBM Carbon Design System |
Vector Databases | 1 server | 14 tools | Milvus, AI/ML Workflows |
Collaboration | 1 server | 27 tools | Atlassian (Jira, Confluence) |
Cloud Storage | 1 server | 42 tools | Box, File Management |
Custom Servers | Unlimited | Custom | Your own integrations |
- π Plug & Play: Connect any MCP-compatible server
- ποΈ Custom Builder: Visual server creation with tool selection
- π Multi-Protocol: HTTP REST APIs + WebSocket/STDIO bridges
- βοΈ Cloud Native: Built for IBM Cloud, AWS, Azure deployment
- π Enterprise Security: OAuth, API keys, role-based access
- π Analytics: Server performance and usage monitoring
Create your own MCP server in minutes:
# 1. Start Fusion MCP Hub
git clone https://github.ibm.com/BPManagementTools/fusion-mcp-hub.git
cd fusion-mcp-hub
npm install
npm run dev
# 2. Open Custom Server Builder
open http://localhost:3000/custom-servers
# 3. Select tools, configure, and download your custom server
# 4. Deploy with dual-mode support (STDIO + HTTP)
Start with included enterprise servers:
# Start all pre-built servers
./start-all-servers.sh
# Or start individual servers
npm run dev # Fusion MCP Hub (port 3000)
node server-salesforce.js # Enterprise CRM tools
node server-milvus.js # Vector database tools
node server-carbon.js # Design system tools
Add any MCP-compatible server:
# Add community servers
python your-custom-mcp-server.py
# Connect to Fusion MCP Hub via HTTP or WebSocket
# Configure in Claude Desktop or your AI client
Maximum Compatibility & Flexibility
{
"mcpServers": {
"fusion-hub": {
"command": "node",
"args": ["http-client.js"],
"env": {
"MCP_SERVER_URL": "http://localhost:3000"
}
}
}
}
{
"mcpServers": {
"fusion-hub": {
"command": "node",
"args": ["stdio-server.js"]
}
}
}
Protocol | Best For | Benefits |
---|---|---|
HTTP | Production, Cloud, New Projects | Simpler deployment, better performance, cloud-native |
STDIO | Development, Legacy, Existing Setups | Protocol compatibility, existing workflows |
Deploy anywhere with enterprise-grade features:
# Kubernetes deployment example
apiVersion: apps/v1
kind: Deployment
metadata:
name: fusion-mcp-hub
spec:
replicas: 3
selector:
matchLabels:
app: fusion-mcp-hub
template:
spec:
containers:
- name: fusion-mcp-hub
image: fusion-mcp-hub:latest
ports:
- containerPort: 3000
Supported Platforms:
- β IBM Cloud (Code Engine, Cloud Foundry)
- β AWS (ECS, Lambda, EC2)
- β Azure (Container Instances, App Service)
- β Google Cloud (Cloud Run, GKE)
- β Kubernetes (Any cluster)
- β Docker (Local or cloud)
- Open
http://localhost:3000/custom-servers
- Select tools from multiple sources
- Configure authentication and settings
- Download your custom server package
- Deploy anywhere
// your-custom-server.js
import { Server } from "@modelcontextprotocol/sdk/server/index.js";
const server = new Server(
{
name: "my-custom-server",
version: "1.0.0",
},
{
capabilities: {
tools: {},
},
}
);
// Add your tools
server.setRequestHandler(ListToolsRequestSchema, async () => ({
tools: [
{
name: "my_tool",
description: "My custom tool",
inputSchema: {
/* your schema */
},
},
],
}));
export default server;
Browse and install from the community marketplace:
- Database connectors (PostgreSQL, MongoDB, Redis)
- API integrations (GitHub, Slack, Discord)
- Development tools (Docker, Kubernetes, CI/CD)
- AI/ML tools (OpenAI, Hugging Face, TensorFlow)
These are just examples of what's possible with Fusion MCP Hub:
- 5 specialized servers with 151+ tools
- Complete API coverage for modern CRM platforms
- Business operations, admin, DevOps, and development workflows
- Design system components and utilities
- Theme management and component libraries
- UI/UX development acceleration
- Vector similarity search and embedding management
- Machine learning workflow integration
- AI-powered data processing and analysis
- Project management and issue tracking
- Knowledge base and documentation management
- Team workflow automation
- Enterprise file management and sharing
- Document processing and collaboration
- Content lifecycle management
Want different integrations? Use the Custom Server Builder to create exactly what you need!
# Copy environment template
cp .env.example .env
# Configure your settings
FUSION_MCP_PORT=3000
FUSION_MCP_HOST=0.0.0.0
# Add your API keys and configurations
OPENAI_API_KEY=your_key_here
# Add other service credentials as needed
{
"mcpServers": {
"fusion-mcp-hub": {
"command": "node",
"args": ["http-client.js"],
"env": {
"MCP_SERVER_URL": "http://localhost:3000"
}
}
}
}
- π Platform Guide - Complete platform documentation
- ποΈ Custom Server Builder - Build your own servers
- βοΈ Deployment Guide - Production deployment
- π API Reference - HTTP and WebSocket APIs
- π§ͺ Testing Guide - Testing your integrations
- Server Marketplace - Share and discover MCP servers
- Tool Library - Reusable tool components
- Templates - Quick-start server templates
- Best Practices - Community-driven guidelines
# Fork and clone
git clone https://github.ibm.com/your-username/fusion-mcp-hub.git
# Create feature branch
git checkout -b feature/your-feature
# Make your changes and test
npm test
npm run lint
# Submit pull request
Ways to Contribute:
- οΏ½οΈ New Server Integrations - Add support for new platforms
- π§ Tool Development - Create reusable tools
- π Documentation - Improve guides and examples
- π Bug Reports - Help us improve reliability
- π‘ Feature Requests - Suggest new capabilities
- Explore the Custom Server Builder
- Connect your existing tools and APIs
- Deploy to your preferred cloud platform
- Share your servers with the community
- Evaluate with included enterprise servers
- Customize for your specific workflows
- Deploy to your infrastructure
- Scale across your organization
- Connect your AI models and tools
- Build intelligent automation workflows
- Experiment with multi-agent systems
- Integrate with popular AI platforms
Ready to transform your development workflow? Start with the Custom Server Builder and create your first MCP server in minutes!
git clone https://github.ibm.com/BPManagementTools/fusion-mcp-hub.git
cd fusion-mcp-hub
npm install && npm run dev
open http://localhost:3000/custom-servers
Complete Salesforce Management Solution
- 59 specialized tools for comprehensive Salesforce operations (now split into role-based servers above)
- Port: 3001
Design System Integration Hub
- Carbon Design System integration for Salesforce components
- Repository Access: Browse and search Carbon component libraries
- Code Discovery: Find implementation patterns and examples
- Metadata Access: Component schemas and documentation
- GitHub Integration: Direct access to Carbon repositories
AI-Powered Code Intelligence
- Vector Database: Semantic code search and discovery
- RAG Tools: Retrieval Augmented Generation for code assistance
- Similarity Search: Find related code patterns and implementations
- Embedding Management: Store and query code embeddings
- Collection Operations: Database management and optimization
- Role-Based Specialization: Connect to only the tools you need for your specific role
- AI-Powered Development: Leverage AI for smarter Salesforce development
- Improved Performance: Smaller, focused servers with faster startup times
- Better Organization: Tools grouped by business function and user role
- Design System Integration: Direct access to Carbon components and patterns
- Semantic Code Search: Find relevant code using natural language queries
- Comprehensive Salesforce Tools: 62+ tools covering all aspects of Salesforce
- Enterprise Ready: Built for IBM Sales Cloud with enterprise-grade features
- Scalable Architecture: Easy to add more specialized servers in the future
- Open Source: All servers are open source and extensible
- Framework: Next.js 15 with Turbopack
- Styling: Tailwind CSS 4
- Animations: Framer Motion
- UI Components: Radix UI
- Icons: Lucide React
- TypeScript: Full type safety
- Font: Inter (Google Fonts)
- Node.js 18+
- npm or yarn
-
Clone the repository:
git clone https://github.com/your-org/fusion-mcp-hub cd fusion-mcp-hub
-
Install dependencies:
npm install
-
Start the development server:
npm run dev
-
Open your browser:
http://localhost:3000
Each MCP server has its own setup requirements. For the specialized Salesforce servers, you'll need to set up the bridge services:
The easiest way to test the MCP servers locally is using the included WebSocket bridge scripts:
-
Install and build all specialized Salesforce servers:
# Clone and setup each server git clone https://github.com/kirtijha/salesforce-admin-mcp-server git clone https://github.com/kirtijha/salesforce-devops-mcp-server git clone https://github.com/kirtijha/salesforce-developer-mcp-server git clone https://github.com/kirtijha/salesforce-business-mcp-server # Build each server cd salesforce-admin-mcp-server && npm install && npm run build && cd .. cd salesforce-devops-mcp-server && npm install && npm run build && cd .. cd salesforce-developer-mcp-server && npm install && npm run build && cd .. cd salesforce-business-mcp-server && npm install && npm run build && cd ..
-
Configure Salesforce credentials (same for all servers):
# Create .env file in each server directory SALESFORCE_USERNAME="your_username" SALESFORCE_PASSWORD="your_password_and_security_token" SALESFORCE_LOGIN_URL="https://login.salesforce.com" # or test.salesforce.com for sandbox
-
Start individual bridge servers:
# From the hub directory, start any bridge: node mcp-bridge-admin.js # Starts Admin server on port 3002 node mcp-bridge-devops.js # Starts DevOps server on port 3003 node mcp-bridge-developer.js # Starts Developer server on port 3004 node mcp-bridge-business.js # Starts Business server on port 3005 node mcp-bridge-carbon.js # Starts Carbon server on port 3009 node mcp-bridge-milvus.js # Starts Milvus server on port 3010 node mcp-bridge-atlassian.js # Starts Atlassian server on port 3011 node mcp-bridge-box.js # Starts Box server on port 3012
Note for External Servers: Before starting the Atlassian and Box bridge servers, ensure their HTTP servers are running:
# Terminal 1: Start Atlassian HTTP server cd ../mcp-atlassian && source .venv/bin/activate && python http_server.py # Terminal 2: Start Box HTTP server cd ../mcp-server-box && source .venv/bin/activate && python http_server.py # Terminal 3: Start bridges (from hub directory) node mcp-bridge-atlassian.js # Connects to http://localhost:3007 node mcp-bridge-box.js # Connects to http://localhost:3008
-
Connect and test in the hub:
- Navigate to any server page in the hub
- The connection form will auto-populate with the correct WebSocket endpoint
- Click "Connect to Server" to establish connection
- Use the "Server Testing" tab to test individual tools
git clone https://github.com/kirtijha/salesforce-admin-mcp-server
cd salesforce-admin-mcp-server
npm install && npm run build
# Configure .env file with Salesforce credentials
# Start bridge: node ../fusion-mcp-hub/mcp-bridge-admin.js
git clone https://github.com/kirtijha/salesforce-devops-mcp-server
cd salesforce-devops-mcp-server
npm install && npm run build
# Configure .env file with Salesforce credentials
# Start bridge: node ../fusion-mcp-hub/mcp-bridge-devops.js
git clone https://github.com/kirtijha/salesforce-developer-mcp-server
cd salesforce-developer-mcp-server
npm install && npm run build
# Configure .env file with Salesforce credentials
# Start bridge: node ../fusion-mcp-hub/mcp-bridge-developer.js
git clone https://github.com/kirtijha/salesforce-business-mcp-server
cd salesforce-business-mcp-server
npm install && npm run build
# Configure .env file with Salesforce credentials
# Start bridge: node ../fusion-mcp-hub/mcp-bridge-business.js
npm install @kirtijha/salesforce-mcp-server
# Configure Salesforce credentials
export SALESFORCE_USERNAME="your_username"
export SALESFORCE_PASSWORD="your_password_and_security_token"
git clone https://github.com/your-org/carbon-salesforce-mcp-server
npm install && npm run build
git clone https://github.com/your-org/mcp-server-isc-milvus
pip install -r requirements.txt
python run_server.py
-
Start the Hub:
cd fusion-mcp-hub npm run dev # Hub runs on http://localhost:3000
-
Start a Bridge Server (example with Admin server):
# In a new terminal cd fusion-mcp-hub node mcp-bridge-admin.js # Bridge runs on ws://localhost:3002
-
Connect in the Hub:
- Navigate to "Salesforce Admin MCP Server" page
- Connection form shows
ws://localhost:3002
(auto-populated) - Click "Connect to Server"
- Status shows "Connected" with green indicator
-
Test Tools:
- Switch to "Server Testing" tab
- Select any tool from the dropdown (e.g., "salesforce_org_limits")
- Fill in required parameters
- Click "Execute Tool" to test with live Salesforce data
Each specialized server runs on a different port:
Server | Default Endpoint | Port | Tools |
---|---|---|---|
Admin | ws://localhost:3002 |
3002 | 16 administrative tools |
DevOps | ws://localhost:3003 |
3003 | 10 deployment tools |
Developer | ws://localhost:3004 |
3004 | 10 development tools |
Business | ws://localhost:3005 |
3005 | 26 business tools |
Legacy | ws://localhost:3001 |
3001 | 59 legacy tools |
Connection Issues:
- Ensure the bridge server is running and showing "WebSocket server listening on port XXXX"
- Check that Salesforce credentials are configured in the server's .env file
- Verify the endpoint URL matches the running bridge port
Tool Execution Issues:
- Check Salesforce credentials and org access
- Verify required parameters are provided correctly
- Check the bridge terminal for detailed error messages
Build Issues:
- Ensure Node.js 18+ is installed
- Run
npm install
in each server directory - Check that TypeScript compilation completes without errors
fusion-mcp-hub/
βββ src/
β βββ app/ # Next.js app router
β β βββ globals.css # Global styles
β β βββ layout.tsx # Root layout
β β βββ page.tsx # Home page
β β βββ servers/ # Server detail pages
β βββ components/ # React components
β β βββ Hero.tsx # Hero section
β β βββ Features.tsx # Features showcase
β β βββ MCPServerCard.tsx # Server card component
β β βββ GetStarted.tsx # Getting started section
β βββ data/
β βββ mcpServers.ts # MCP server data and types
βββ public/ # Static assets
βββ package.json
βββ README.md
Create personalized MCP servers by combining tools from multiple sources:
- Tool Selection: Pick and choose from 238+ available tools across all servers
- Dual Transport: Generated servers support both STDIO and HTTP protocols
- Production Ready: Includes health checks, monitoring, and error handling
- VS Code Integration: Auto-generated configuration files for immediate use
- Comprehensive Documentation: Each server comes with setup guides and API docs
- Select Tools: Browse and select tools from Salesforce, Milvus, Box, and Atlassian servers
- Configure Server: Set name, description, port, and deployment options
- Generate & Download: Get a complete, ready-to-deploy MCP server package
- Deploy: Use STDIO mode for VS Code or HTTP mode for web applications
- Salesforce Developer Kit: Combine Salesforce SOQL, Apex, and metadata tools
- Content Management Hub: Mix Box storage with Confluence documentation tools
- Data Analytics Server: Combine Milvus vector search with Salesforce reporting
- DevOps Pipeline: Merge Salesforce deployment with Jira tracking tools
π Try the Custom Server Builder
- Responsive Design: Works on all devices and screen sizes
- Smooth Animations: Framer Motion powered transitions
- Modern Design: Clean, professional interface with gradients
- Accessible: Built with accessibility best practices
- Tool Documentation: Detailed information about each MCP tool
- Setup Guides: Step-by-step installation instructions
- Use Cases: Real-world application examples
- Code Examples: Practical implementation snippets
- Expandable Sections: Accordion-style tool and setup information
- Visual Hierarchy: Clear organization of information
- Interactive Cards: Hover effects and smooth transitions
- Type Safety: Full TypeScript implementation
This hub showcases a modern approach to AI-powered development tooling:
- Model Context Protocol (MCP): Standard interface for AI tool integration
- Modular Design: Each server focuses on specific capabilities
- Comprehensive Coverage: From low-level data operations to high-level workflows
- Enterprise Integration: Built for real-world Salesforce development scenarios
Each MCP server includes:
- Comprehensive tool documentation
- Setup and configuration guides
- Code examples and use cases
- API references and schemas
- Troubleshooting guides
npm run build
# Deploy to Vercel
vercel --prod
# Build the application
npm run build
# Create Docker image
docker build -t mcp-hub .
# Run container
docker run -p 3000:3000 mcp-hub
These MCP servers are open source and welcome contributions:
- Bug reports and feature requests
- Code contributions and improvements
- Documentation enhancements
- Community examples and patterns
For questions, issues, or support:
- Check individual server documentation
- Open GitHub issues in respective repositories
This project is licensed under the MIT License - see the LICENSE file for details.
Built with β€οΈ for developers by @kirtijha