Labby

Local-first MCP gateway and homelab control plane with CLI, HTTP API, and web UI.

Labby is an MCP server that acts as a local-first gateway and control plane for homelab environments. It is published by ai.dinglebear and available as the npm package labby-mcp at version 1.2.0. The server communicates over stdio transport and exposes its functionality through three interfaces: a command-line interface (CLI), an HTTP API, and a web UI, giving operators multiple ways to interact with the same underlying control plane.

Three environment variables shape its runtime behavior: LABBY_VERSION pins or communicates the active version, LABBY_REPO points to the relevant repository context, and LABBY_LOG controls logging output. These allow the server to be tuned for different deployment scenarios within a local or self-hosted network.

The “local-first” framing means the server is intended to run on infrastructure the developer owns or controls rather than routing through a third-party cloud service. This makes it suited to homelab setups where an agent needs a single coordination point across self-hosted services.

No tool definitions are exposed in the current registry record, so the specific actions an agent can invoke programmatically through the MCP protocol are not documented in available facts. Developers evaluating Labby for agent-driven automation should verify the available tool surface directly from the repository at github.com/jmagar/labby before integrating it into a workflow.

Use cases

  • Use it as a central MCP gateway when coordinating multiple self-hosted services in a homelab
  • Use it to manage homelab control-plane operations through a web UI without writing additional tooling
  • Use it to expose homelab management over an HTTP API consumable by other local services
  • Use it to adjust runtime behavior via environment variables for different local deployment contexts

When to use it

  • When running a homelab and needing a unified control plane accessible via CLI, HTTP API, or web UI
  • When the agent deployment must stay entirely on local, self-owned infrastructure
  • When stdio transport is compatible with the agent framework in use

When not to use it

  • When a cloud-hosted or managed MCP endpoint is required — Labby is local-first and not designed for cloud deployment
  • When a specific, documented tool list is required before integration — no tools are listed in the current registry record
  • When the environment does not support npm package installation
  • When a license-reviewed dependency is mandatory — no license is recorded for this package