RuView ships USB onboarding so a sensing node sets itself up

What RuView is

RuView reads the WiFi signals already bouncing around a room and turns them into presence, movement and room level activity. No cameras, no wearables, and it works in the dark and through walls. It is physics rather than imaging.

It speaks to the common home platforms directly: Home Assistant over MQTT, Apple Home as a HomeKit bridge, and Google Home and Alexa through the same bridge or a Matter endpoint.

What changed in v2655

The release merges pull request #1902, which puts secure USB onboarding on the main firmware line and fixes the ESP32-C6 console receive path that had been going silent. Firmware version 0.8.12.

Onboarding over USB matters because the alternative is the usual captive portal dance. Plug the node into a machine, provision it, unplug it. A related fix keeps the onboarding digest compatible with ESP-IDF 6.

Get started

Node.js 20 or newer. Start with the doctor command, which checks the environment:

npx @ruvnet/ruview@0.4.0 doctor

Expected result: a report of what is present, what is missing and what to install next.

RuView exposes an MCP server for agent hosts:

npx @ruvnet/ruview@0.4.0 mcp start

There is also a container image for the sensing service:

docker run -p 3000:3000 ruvnet/wifi-densepose:latest

Vital signal work in this project is research. Treat breathing and heart rate output as experimental measurement, not a validated medical capability.

Use it today

Practical case: room level occupancy for home automation without putting a camera in the room. Input is the WiFi radio you already have; the workflow is onboard a node, point it at Home Assistant, and expose presence per room; the output is an entity that lights, HVAC or alerts can react to.

Acceptance test: run doctor, confirm it reports a healthy environment, then walk out of the sensed room and watch the presence entity go false within its stated window.

Push it further

Experimental commentary. The ambitious version is a building that understands itself: many cheap nodes, one spatial model, and agents that act on the room rather than on a device list. The limitation is calibration — sensing quality depends heavily on placement and the radio environment.

Falsifiable test: move a node two metres and re-run the same presence walkthrough. If detection quality shifts materially, calibration is a first class part of your deployment and not an afterthought.