Hub MCP Server

Coordinate multiple MCP servers via a unified REST API and single endpoint hub.

Data last scanned yesterday · Reviewed 3 weeks ago

52★ 515JavaScriptMIT

Overview

MCP Hub centralizes your Model Context Protocol environment by acting as a management layer for multiple servers. It provides a REST API and web UI to organize your tools and connections, allowing you to expose all configured server capabilities through a single endpoint. This approach simplifies client configuration, as you only need to point your AI tools to the Hub to access all available resources. The system automatically handles request routing, capability namespacing to prevent conflicts, and provides real-time updates for tools, resources, and prompts. It supports both local and remote transport protocols—including STDIO, SSE, and streamable-http—while offering built-in development features like hot-reloading for server configurations.

Our verdict

With a security score of 42/100, this tool is categorized as high risk. The assessment is driven by the fact that the project is community-maintained rather than vendor-official and has not undergone editorial review. Furthermore, the repository shows limited maintenance with the last commit occurring nearly 300 days ago. Users should exercise caution and review the codebase before integration, particularly given the lack of verified authentication and read-only mode security checks.

Supports namespace isolation for tool requests
Implements the 2025-03-26 MCP specification
Features a minimal footprint with only one direct dependency
Includes auto-reconnection with backoff logic

High security risk score of 42/100
Inactive repository with no commits in 293 days
Missing support for roots and sampling capabilities
Lack of third-party editorial review

Setup

Claude Desktop

{
  "mcpServers": {
    "mcp-hub": {
      "args": [
        "-y",
        "mcp-hub"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›Does MCP Hub support all MCP protocol features?

No, it explicitly does not support Roots, Sampling, or Completion capabilities.

›How do I configure my MCP clients to use the hub?

You only need to configure your clients to connect to the single endpoint at http://localhost:37373/mcp.

›What types of MCP server transports can the hub manage?

It supports STDIO for local servers and both streamable-http and SSE for remote servers.

›Can I use environment variables in my configuration files?

Yes, the hub supports ${} syntax for environment variables and command execution across all configuration fields.

›What authentication methods are supported for connected servers?

The hub supports OAuth 2.0 with PKCE flow and header-based authentication for API keys or tokens.

›Does the hub handle potential tool or resource name collisions?

Yes, the hub automatically namespaces capabilities from different servers to prevent conflicts, such as distinguishing filesystem__search from database__search.