RuView

Name: RuView
Author: rUv

Operate a RuView WiFi-sensing setup: ESP32 CSI nodes, radar, LiDAR, flashing, calibration and training

Official · rUv TypeScript Local · stdio Developer tools DevOps and monitoring

GitHub npm Website

51 popularity score ★ 103k ↓ 460/wk

Connect your client

Local (npm)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server.

Or add it by command or config

Claude CodeTerminal

claude mcp add ruview -- npx -y @ruvnet/ruview mcp start

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "ruview": {
      "command": "npx",
      "args": [
        "-y",
        "@ruvnet/ruview",
        "mcp",
        "start"
      ]
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "ruview": {
      "command": "npx",
      "args": [
        "-y",
        "@ruvnet/ruview",
        "mcp",
        "start"
      ]
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "ruview": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@ruvnet/ruview",
        "mcp",
        "start"
      ]
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.ruview]
command = "npx"
args = ["-y", "@ruvnet/ruview", "mcp", "start"]

Gemini CLITerminal

gemini mcp add --scope user ruview npx -- -y @ruvnet/ruview mcp start

Settings 5

`RUVIEW_MCP_GRANTS`	Optional: Grants enabled at start: device-access, remote-host, workspace-write, hardware-write, credential-use.
`RUVIEW_MCP_TOKEN`	Optional: Bearer token for the HTTP transport (mcp start --http).
`RUVIEW_CREDENTIALS_PATH`	Optional: Non-default Cognitum Spaces credential store for the server.
`RUVIEW_LIDAR_TOKEN`	Optional: Token for the iPhone LiDAR relay WebSocket.
`LIBTORCH`	Optional: Path to libtorch for pose training; or set LIBTORCH_USE_PYTORCH=1.

What it does

RuView is a project that estimates presence, motion, breathing and coarse body pose from WiFi Channel State Information (CSI) collected by cheap ESP32 boards, optionally combined with mmWave radar and LiDAR. The @ruvnet/ruview package is its operator toolkit: the same set of ruview_* functions is available as a command-line program, a JavaScript SDK and an MCP server, so an assistant can help set up and run a sensing installation.

Through the MCP server an assistant can walk a newcomer through choosing a setup path, list serial ports and attached sensors, check that CSI packets are arriving from ESP32 nodes, read 60 GHz and 24 GHz radar modules and RPLIDAR or iPhone LiDAR, plan and perform firmware flashing with checksum and chip checks, run the room calibration pipeline, plan and launch pose or room model training, and gate evaluation results against a mean-pose baseline. It also offers diagnostics of the local toolchain, source-cited guidance about the RuView codebase, a search over a reviewed set of project facts, read-only tool runs on SSH hosts, and paged reads of a Cognitum Spaces account.

A distinctive part is the accuracy guardrail: ruview_claim_check flags accuracy figures in text that are not tagged as measured against a baseline, and the training gate returns the only claim wording it considers publishable. Tools that modify hardware or the workspace are off by default and must be enabled with grants when the server is started and confirmed per call.

It runs over stdio for local clients, or over a loopback HTTP transport with a bearer token for ChatGPT and other MCP Apps hosts, where three tools render in a small console widget.

What it needs

Runtime	Node.js 20+

Node.js 20 or newer. Most useful on the machine the sensors are plugged into: ESP32 CSI nodes, MR60BHA2 or LD2410 radar, or RPLIDAR. Flashing and port listing need Python with esptool and pyserial. Training needs a RuView repository checkout with Rust and libtorch (torch 2.11.0 for tch 0.24). Cognitum Spaces reads need an installed wifi-densepose binary signed in with OAuth.

Limits

Its makers say WiFi sensing gives only coarse pose, presence and breathing and is not a camera; device-reported vitals are labelled unvalidated and the smoke training run and kernel self-test are synthetic. Flashing, calibration and training are never allowed over the HTTP transport, which serves read tools only. A cloud-hosted agent cannot reach local USB ports. The MCP doctor cannot probe a board or contact a network host. Remote SSH hosts can only be added from the CLI and only run read-only tools. Spaces reads are limited to 100 items per page and grant no write authority.

Who it suits

Suits people building or running a RuView WiFi-sensing installation who want an assistant to inspect nodes, read sensors, flash ESP32 firmware or run calibration and training from a local session, with explicit grants guarding every change. It is tied to RuView hardware and its repository, so it has little use without ESP32 CSI nodes or a RuView checkout. Pick the HTTP mode only for read-only monitoring from ChatGPT; anything that writes needs the stdio server on the machine with the hardware.

Tools 22

ruview_onboard: Choose a setup path (Docker demo, repo build or live ESP32) and get the next command
ruview_claim_check: Flag accuracy claims in text that are untagged or overstated
ruview_verify: Run the repository's deterministic proof script and report its verdict
ruview_node_monitor: Check that CSI data is arriving from an ESP32 node
ruview_calibrate: Run the room calibration pipeline from baseline to room watch; needs a write grant
ruview_node_flash: Flash ESP32 firmware with checksum and chip checks and capture the boot log; needs a hardware grant
ruview_firmware_plan: Check a firmware bundle and return the esptool command without running it
ruview_firmware_ports: List serial ports on the machine
ruview_doctor: Report diagnostics for runtime, repo, Rust, Python, esptool, bundles, serial and kernel
ruview_train: Run pose or room model training in the checkout, writing checkpoints; needs a write grant
ruview_train_plan: Work out a training command without running it
ruview_train_gate: Test an evaluation result against a mean-pose baseline and leakage rules
ruview_devices_scan: Identify attached USB serial devices by vendor and product ID
ruview_esp32_capture: Receive CSI, vitals and feature packets from ESP32 nodes over UDP
ruview_mmwave_read: Read a 60 GHz MR60BHA2 or 24 GHz LD2410 radar module over serial
ruview_lidar_read: Summarise a scan from RPLIDAR or an iPhone LiDAR relay
ruview_host_list: List the SSH hosts configured for remote runs
ruview_host_run: Run one read-only tool on a configured SSH host
ruview_guidance: Return source-cited notes on the RuView codebase, maturity and known limits
ruview_spaces_list: Page through Cognitum Spaces sites, zones, entities, events or alerts
ruview_memory_search: Search the project's reviewed, source-cited fact store
ruview_kernel_selftest: Run a synthetic self-test of the optional ruview-kernel compute package

Popularity

51 popularity score, on our list (the bar is 36)

★ 103k ↓ 460/wk

GitHub stars, whole repository	102,785	21.3 points
npm downloads, last week	460	21.3 points
Official server	Yes	8 points

It lives in the folder harness/ruview of ruvnet/ruview, which holds more than this server. Its stars are the whole repository's, so in the score they never count for more than its own downloads.

Measured 2026-11-07. How the score works.

Where it is published

Code	ruvnet/ruview/harness/ruview ★ 103k, the whole repository
npm	@ruvnet/ruview 0.9.3, 2026-11-06

What it can reach

Runs	On your machine
Reaches	Files on your machine, The internet, Runs shell commands
Changes things	Yes: it can write, send, create or delete
Sign-in	None
Licence	MIT (npm)
Last code change	2026-11-06

Runs locally under Node and calls Python esptool, Cargo and SSH as child processes. Read-only tools need no grant; sensor reads need device-access, SSH runs need remote-host, calibration and training need workspace-write and flashing needs hardware-write, each set in RUVIEW_MCP_GRANTS at start, with confirm: true per call for writes. Flashing overwrites ESP32 firmware; training writes checkpoints into the checkout. The HTTP mode binds to loopback, needs a bearer token and refuses write grants; for ChatGPT the token goes in the connector URL. SSH hosts come from ~/.config/ruview/hosts.json with strict host-key checking. Spaces reads use the wifi-densepose OAuth session, need credential-use, and only reach api.cognitum.one.

Over time

GitHub stars, whole repository 103k +5.0k (+5 %)

103k98k

13 Oct 20267 Nov 2026

npm downloads a week 460 +430 (+1433 %)

72530

29 Jun 20267 Nov 2026

Before the site measured it, read from npm's daily download counts.

Popularity score 51 −30 (−37 %)

8133

13 Oct 20267 Nov 2026

What is rising across the directory

Sources

Page checked 2026-10-14.

Badge for your README

Shows this server's rank in its category, and stars for a server that runs on your machine. It updates every hour.

MarkdownREADME.md

[![RuView on EveryMCPServer](https://everymcpserver.com/badge/ruview.svg)](https://everymcpserver.com/servers/ruview)

Connect your client

Add in one click

Or add it by command or config

Settings 5

What it does

What it needs

Limits

Who it suits

Tools 22

Similar servers

ruv-swarm

RuVector MCP Server

Omada MCP

Sentry MCP

Ruflo

RuvNet Brain