Overview

MCPHub is an open-source gateway for MCP servers. It gives AI applications a single place to connect, manage, and organize their tools, so developers can spend more time building the application itself.

Who it's for

AI developers connecting multiple MCP servers, and teams that want a shared place to manage their agents' tools.

Use it when

You need a unified endpoint for multiple MCP servers.
Different applications need different groups of tools or routing rules.
You want to manage access and keep an eye on server health.
Your tool collection has grown beyond a handful of individual configurations.

Key features

Unified gateway: aggregate MCP servers behind a shared endpoint.
Flexible organization: use groups and routing to expose the tools a client needs.
Management interface: configure servers and inspect their status in one place.
Open source and self-hosted: run it in your own environment and contribute improvements.

Why I built it

As AI applications connect to more tools, managing each server separately becomes repetitive work. MCPHub moves that work into a shared management layer, making tool integration easier to maintain over time.

Compared with direct connections

Connecting directly to a single MCP server works well for simple setups. MCPHub is useful when several clients need to reuse several servers, with shared configuration and a central management interface. It works alongside the underlying MCP servers.

Status

Active and open source. See the documentation and GitHub releases for current capabilities, deployment instructions, and version changes.

FAQ

Is MCPHub an AI model?

MCPHub is infrastructure for connecting and managing MCP servers. It works with AI clients that support MCP.

Can I host it myself?

Yes. The project is open source. Deployment instructions and configuration references are available in the documentation.