What is this?
RuView turns ordinary WiFi signals into a spatial-intelligence sensor — it detects presence, measures breathing and heart rate, tracks motion and falls, and even estimates body pose through walls, using Channel State Information (CSI) captured by cheap ESP32 boards, with no cameras or wearables.
RuView turns ordinary WiFi signals into a spatial-intelligence sensor — it detects presence, measures breathing and heart rate, tracks motion and falls, and even estimates body pose through walls, using Channel State Information (CSI) captured by cheap ESP32 boards, with no cameras or wearables.Is this one app, or a whole platform?It is a full sensing platform, not a single app: ESP32 firmware captures CSI, a Rust server processes and serves it, an npm-published MetaHarness runs AI-guided onboarding/calibration, and a catalog of over 100 modular 'Cogs' adds specific capabilities like pose estimation or fall detection.
It is a full sensing platform, not a single app: ESP32 firmware captures CSI, a Rust server processes and serves it, an npm-published MetaHarness runs AI-guided onboarding/calibration, and a catalog of over 100 modular 'Cogs' adds specific capabilities like pose estimation or fall detection.How do I get it running?There are four quick-start paths: run the Docker image with simulated data for evaluation, flash an ESP32-S3/C6 node for real CSI capture, pair with a Cognitum Seed for persistent memory, or just verify the signal pipeline in Python with zero hardware.
There are four quick-start paths: run the Docker image with simulated data for evaluation, flash an ESP32-S3/C6 node for real CSI capture, pair with a Cognitum Seed for persistent memory, or just verify the signal pipeline in Python with zero hardware.Do I need special WiFi hardware to use it?For real sensing, yes: presence, vitals, through-wall detection and pose estimation all require Channel State Information (CSI), which comes from an ESP32-S3 (~$9) or a research NIC; a plain WiFi laptop only gives coarse RSSI-based presence detection.
For real sensing, yes: presence, vitals, through-wall detection and pose estimation all require Channel State Information (CSI), which comes from an ESP32-S3 (~$9) or a research NIC; a plain WiFi laptop only gives coarse RSSI-based presence detection.How accurate is the detection right now?The presence/vitals model is honestly re-benchmarked at 82.3% held-out accuracy, correcting an earlier retracted '100%' claim that was measured on a single-class recording; the on-device 17-keypoint pose model is explicitly labeled a 'first-cut' with only 3.0% PCK@20, well below the project's own 35% target.
The presence/vitals model is honestly re-benchmarked at 82.3% held-out accuracy, correcting an earlier retracted '100%' claim that was measured on a single-class recording; the on-device 17-keypoint pose model is explicitly labeled a 'first-cut' with only 3.0% PCK@20, well below the project's own 35% target.Does any of my sensing data go to the cloud?No — RuView runs entirely on edge hardware (the ESP32 mesh plus an optional Cognitum Seed) with no cloud, no cameras, and no internet connection required; sensing events can carry attached privacy-policy and provenance metadata.
No — RuView runs entirely on edge hardware (the ESP32 mesh plus an optional Cognitum Seed) with no cloud, no cameras, and no internet connection required; sensing events can carry attached privacy-policy and provenance metadata.What is the Cognitum Seed?It is an optional edge companion device that pairs with the ESP32 mesh to add a persistent vector-store memory, kNN search, cryptographic Ed25519 witness-chain attestation, and AI integration, for about $140 total hardware cost.
It is an optional edge companion device that pairs with the ESP32 mesh to add a persistent vector-store memory, kNN search, cryptographic Ed25519 witness-chain attestation, and AI integration, for about $140 total hardware cost.What is HOMECORE?HOMECORE is RuView's local automation layer — it tracks state and history, runs automations and signed Wasm plugins, and exposes voice hooks and HomeKit support, all without any cloud dependency.
HOMECORE is RuView's local automation layer — it tracks state and history, runs automations and signed Wasm plugins, and exposes voice hooks and HomeKit support, all without any cloud dependency.What is the RuView MetaHarness?It is an npm package (@ruvnet/ruview) acting as an AI operator — a guarded Claude Code/Codex agent that helps onboard, calibrate, train, and verify sensing claims, running read-only by default unless explicitly granted write access.
It is an npm package (@ruvnet/ruview) acting as an AI operator — a guarded Claude Code/Codex agent that helps onboard, calibrate, train, and verify sensing claims, running read-only by default unless explicitly granted write access.Which smart-home systems does it integrate with?It works natively with Home Assistant via an MQTT publisher, Apple Home/HomePod as a HAP-1.1 bridge, and Google Home/Amazon Alexa via the same bridge or a Matter endpoint, exposing 21 entities per sensor node.
It works natively with Home Assistant via an MQTT publisher, Apple Home/HomePod as a HAP-1.1 bridge, and Google Home/Amazon Alexa via the same bridge or a Matter endpoint, exposing 21 entities per sensor node.What are similar projects?papermono-shopping-list, esphome, techo5, muse-gadget-sdk, esp32sim, voiceshell-muse-bridge, oomwoo, kindle-wall-display
papermono-shopping-list, esphome, techo5, muse-gadget-sdk, esp32sim, voiceshell-muse-bridge, oomwoo, kindle-wall-displayWhich topics is it in?self-learning, wifi-security, pose-estimation, home-assistant, wifi, home-automation, esp32, embedded-systems, firmware, iot, raspberry-pi, hardware
self-learning, wifi-security, pose-estimation, home-assistant, wifi, home-automation, esp32, embedded-systems, firmware, iot, raspberry-pi, hardware