A Kubernetes control plane for your Model Context Protocol sprawl

What it does MCPCAN is a web-based ops platform for managing Model Context Protocol (MCP) services. It containerizes each MCP server, handles token-based auth, and exposes them over SSE, STDIO, or STREAMABLEHTTP. The stack is Go microservices on the backend, Vue 3 on the frontend, with MySQL and Redis underneath.

The interesting bit The protocol conversion layer is the quiet workhorse. MCP servers speak different transports; MCPCAN normalizes them so your Dify workflow or OpenAI integration doesn’t need to care whether the underlying server was built for SSE or STDIO. That’s the kind of glue that sounds boring until you’re debugging three conflicting Python environments at 2 AM.

Key highlights

• Ships as Docker Compose or Helm charts for Kubernetes
• RBAC and department-level permission management via the MCP-Authz service
• Real-time container monitoring baked into the dashboard
• Enterprise features live in .ee folders under a separate commercial license
• Demo runs at demo.mcpcan.com with admin/admin123 credentials (not a recommendation, just a fact)

Caveats

• The “Sustainable Use License” is not OSI-approved; the .ee split suggests open-core positioning that isn’t fully explained in the README
• README repeats itself almost verbatim in the Introduction and first paragraph, which hints at documentation still finding its footing

Verdict Worth a look if you’re running more than a handful of MCP servers in production and need visibility, auth, and rollback. Skip it if you’re just experimenting with a single local MCP server; the overhead is overkill for one Python script.