mcp-hub
A centralized manager for Model Context Protocol (MCP) servers with dynamic server management and monitoring
claude mcp add --transport stdio ravitemer-mcp-hub npx -y mcp-hub --port 37373 --config path/to/config.json
How to use
MCP Hub acts as a central coordinator that lets you manage multiple MCP servers and expose a single client endpoint for MCP capabilities. It provides a REST API under /api for server management and a unified MCP endpoint at /mcp for MCP clients to access tools, resources, prompts, and templates across all connected servers. Clients such as Claude Desktop or other MCP clients can connect to localhost:37373/mcp and discover capabilities from every registered MCP server, with real-time updates and automatic namespacing to avoid conflicts. The hub supports OAuth with PKCE, header-based authentication, and dynamic server management including starting, stopping, and auto-reconnection when servers change. It also offers a VS Code-friendly configuration format and JSON5 support for comment-friendly config files, enabling you to define servers with environment variable substitutions and command executions across all fields.
How to install
Prerequisites:
- Node.js (recommended: latest LTS) and npm
- Access to install global npm packages
Install MCP Hub globally:
npm install -g mcp-hub
Basic usage to start the hub:
mcp-hub --port 3000 --config /path/to/config.json
If you prefer to use a local config, you can also run with a merged set of config files:
mcp-hub --port 3000 --config ~/.config/mcphub/global.json --config ./.mcphub/project.json
Configuration files can be authored using the MCP Hub JSON format (with ${} style substitutions). See the README for examples of mcpServers configuration and VS Code compatibility. Ensure your config enables the servers you want to manage and uses the universal placeholder syntax for environment variables and command execution.
Additional notes
Tips and notes:
- MCP Hub routes capabilities from all managed servers to a single /mcp endpoint. Use /api for server management and status events via SSE.
- Use ${env:VAR} or ${VAR} in config values to inject environment variables, and leverage ${cmd: ...} for command execution within config fields.
- For client configuration, you can use a simplified endpoint example like {"mcpServers": {"Hub": {"url": "http://localhost:37373/mcp"}}} to point clients at the hub.
- If you rely on VS Code-style configuration, MCP Hub supports both mcpServers and servers keys and can consume JSON5 with trailing commas or comments.
- For authentication, you can enable OAuth 2.0 with PKCE and header-based tokens; consider setting up API keys in headers for secure access.
- If you encounter issues starting the hub, check that the port is available, the config file path is correct, and that the environment variables referenced in the config are exported in your shell.
- The hub provides real-time capability updates and server status through SSE; monitor /api/events for events and state changes.
Related MCP Servers
mcp-agent
Build effective agents using Model Context Protocol and simple workflow patterns
easy
Absurdly easy Model Context Protocol Servers in Typescript
mcp -odoo
A Model Context Protocol (MCP) server that enables AI assistants to securely interact with Odoo ERP systems through standardized resources and tools for data retrieval and manipulation.
openapi
OpenAPI definitions, converters and LLM function calling schema composer.
mxcp
Model eXecution + Context Protocol: Enterprise-Grade Data-to-AI Infrastructure
mcp-chain-of-draft
Chain of Draft Server is a powerful AI-driven tool that helps developers make better decisions through systematic, iterative refinement of thoughts and designs. It integrates seamlessly with popular AI agents and provides a structured approach to reasoning, API design, architecture decisions, code reviews, and implementation planning.