2.3 Useful CLI commands | Installation

These commands help you check the MCP and renderer status, and stop services when you are done.

To start a remote MCP that can be shared with a team, use this command:

luminon start remote

Remote MCP started (pid 4938). Log: /Users/luminon/Documents/luminon/logs/remote.log Remote MCP: http://localhost:3001/mcp Health: http://localhost:3001/health

When you want to check the current state, run `luminon status`:

luminon status

Data dir: /Users/luminon/Documents/luminon Lock file: /Users/luminon/Documents/luminon/.luminon-lock.json MCP: host-managed or stopped (AI Tool stdio MCP may not appear here) Renderer: running (pid 4615, log /Users/luminon/Documents/luminon/logs/renderer.log) Remote MCP: running (pid 4938, log /Users/luminon/Documents/luminon/logs/remote.log)

To stop services separately, use `luminon stop renderer` to close the renderer and `luminon stop remote` to shut down the remote MCP. If you try to start a process that is already running, Luminon detects it and shows a message like `Renderer is already running. Use 'luminon status' or 'luminon stop renderer'.`

luminon stop renderer

luminon stop remote

For authenticated remote MCP access, generate tokens first. You can create multiple tokens and use them by team, service, or environment (for example: marketing, BI, staging, production). `luminon start remote` uses the last created token if one exists, and `luminon token current` shows which token the remote MCP is currently using.

luminon token create marketing-team-token

Token created: id=d050fba85a6d name=marketing-team-token Bearer token (save now, shown once): lum_V2-ANwXj0Ag10DckKFqb6bTkZ-7OTp2p

luminon token list

173d265a8803 name=token-test created=2026-05-20T14:21:32.003Z d050fba85a6d name=marketing-team-token (default) created=2026-06-01T16:49:17.910Z

luminon token delete marketing-team-token

Deleted token 'marketing-team-token'.

luminon token current

173d265a8803 name=token-test created=2026-05-20T14:21:32.003Z

Next, we will configure Luminon with the main AI tools. You can also use it with any MCP client, including a local model, if you prefer not to expose your data or pay for token usage.