Base
Hardened build
minimal ports, security patches applied at build time
MCPHub
Azure
Artificial Intelligence (AI)
1 variation
A self hosted gateway for the Model Context Protocol: connect MCP servers once, then serve every server, each group or a single server to AI clients through authenticated endpoints
Variations
This product is available in the build below. Open it for the operating system, cloud and version detail, or read its deployment guide.
| Product | Cloud | Operating system | Version | |
|---|---|---|---|---|
| MCPHub on Ubuntu 24.04 | Azure | Ubuntu 24.04 | Standard | View · Guide |
Overview
MCPHub is an open source gateway and control plane for the Model Context Protocol, the standard that lets AI assistants and agents call external tools. You connect MCP servers to it once, local stdio servers started with npx or uvx as well as remote SSE and streamable HTTP servers, and MCPHub publishes them to clients such as Claude Code, Cursor and OpenWebUI through stable endpoints: one for every server, one for each group and one for each single server. Groups can expose only chosen tools from a server, individual tools can be switched off or have their descriptions rewritten, and bearer keys can be limited to particular groups or servers, so each agent sees exactly the tools it needs. A web dashboard manages servers, groups, users, keys and per user credentials, with logs and health status, and it all runs as one lightweight Node.js process with a JSON settings file.
Why the cloudimg image
Out of the box upstream MCPHub creates its admin account with a random password that it prints to the server log. cloudimg generates the admin password, the MCP bearer key, the session signing secret and the credential encryption key for each virtual machine at first boot, passes the password to MCPHub from a root only file so it never reaches a log, and proves the admin login and that every MCP endpoint refuses callers without a key before nginx is allowed to start. MCPHub runs as an unprivileged account on loopback only behind nginx with HTTPS and a certificate generated per virtual machine, and port 80 only redirects. It is built from the upstream release tag with the upstream lockfile, ships two example MCP servers that need no network or API key, and every deployment is fully patched with a paired deploy guide and 24/7 cloudimg support.
Common uses
- Give Claude Code, Cursor and other AI clients one authenticated MCP endpoint for all of a team's tools
- Expose a chosen subset of tools to each agent through its own group endpoint and scoped key
- Run local npx and uvx MCP servers centrally instead of on every developer machine
More in Artificial Intelligence (AI)
Argilla
Artificial Intelligence (AI)
Argilla, an open source data annotation and curation platform for building high quality datasets for natural language pr
Bifrost
Artificial Intelligence (AI)
one OpenAI compatible endpoint in front of every model provider, packaged by cloudimg
ContextForge MCP Gateway
Artificial Intelligence (AI)
ContextForge, the open source model context protocol gateway from IBM: federate many MCP servers behind one authenticate
Crawl4AI
Artificial Intelligence (AI)
an open source web crawler that turns any web page into clean, LLM ready markdown over a REST API, secured from first bo
Do
DeepWiki-Open
Artificial Intelligence (AI)
turn any Git repository into a browsable, AI generated wiki
Dify
Artificial Intelligence (AI)
Dify, a source available platform for building LLM apps, agents, RAG and workflows from a visual console.