Mcphub MCP Server

Unified management and routing for multiple MCP servers via HTTP/SSE endpoints.

Data last scanned today · Reviewed today

58★ 2,239TypeScriptApache-2.0

Overview

MCPHub acts as a centralized dashboard and proxy to organize and scale MCP server deployments. It enables flexible routing by allowing users to access servers individually, in groups, or collectively, providing granular control over the visibility of tools, prompts, and resources for each connected instance. The system includes smart features such as AI-powered tool discovery via vector semantic search and transparent result compression to manage large output streams. It supports dynamic configuration updates without downtime and offers native OAuth 2.0 and social login integration to handle authentication across diverse server environments.

Our verdict

With a security score of 48/100, this project is classified as high risk. The score reflects that the project is community-maintained rather than vendor-official and its core security components, including OAuth support and read-only mode, have not yet been editorially reviewed. Due to the high-risk classification and the lack of external security vetting, it should be treated with caution and is recommended primarily for users comfortable with self-managed open-source infrastructure.

Supports hot-swappable configurations without requiring downtime
Provides built-in AI-powered tool discovery via semantic search
Maintained by 35 contributors with daily commit activity
Includes Docker-ready images for streamlined containerized deployment

Authentication methods have not yet been editorially reviewed
Core security features like read-only mode lack independent verification
Lacks backing from an official vendor or organization

Setup

Claude Desktop

{
  "mcpServers": {
    "mcphub": {
      "args": [
        "-y",
        "@samanhappy/mcphub"
      ],
      "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

›What is the difference between the 'latest' and 'latest-full' Docker images?

The 'latest' image includes essential tools like Node.js, Python, and uv, while the 'latest-full' image adds the Rust toolchain, Docker Engine, and Playwright browsers for more demanding environments.

›How can I manage MCP servers without using the web dashboard?

You can set the DISABLE_WEB environment variable to true to run the hub in headless mode, or use the built-in CLI to list, add, and call tools from your terminal.

›Does MCPHub support persistent storage for configurations?

Yes, you can enable Database Mode using PostgreSQL for production environments, or simply mount a local data directory in your Docker container to preserve state and credentials.

›How do I secure my MCP endpoints?

MCP endpoints require bearer authentication by default to prevent accidental exposure. You can toggle this setting in the Keys section of the dashboard.

›How is the administrator password handled on the first launch?

If you do not specify an ADMIN_PASSWORD environment variable, a random password is generated and printed directly to the server logs upon first launch.