Ruview

What @ruvnet/ruview does

@ruvnet/ruview is the operator layer for RuView’s sensing platform. It connects an AI agent to host-device workflows involving ESP32 Channel State Information nodes, radar, and LiDAR. The package is designed for operational tasks rather than only passive documentation: an agent can inspect the local setup, request source-cited guidance, observe connected hardware, and work through sensing setup procedures.

The harness also covers remote hosts over SSH, diagnostics, verified ESP32 firmware flashing, calibration, evidence-gated training, deterministic verification, and Cognitum Spaces. These controls are intended to keep sensing claims tied to evidence and to make device operations reviewable.

How it works

The package exposes a CLI, SDK, and MCP server. The MCP endpoint can run over standard input/output or HTTP, allowing an MCP client to invoke the operator workflow while the underlying process accesses the host and attached sensing equipment. The README describes one MCP server for agent use and a live console widget, along with guarded Claude Code and Codex agents.

RuView’s sensing inputs include CSI from ESP32 nodes and other supported hardware such as 60 GHz radar and LiDAR. The broader platform can process radio observations for presence, movement, breathing, heart rate, activity, and related room-sensing tasks, but the harness’s role is to help onboard, inspect, calibrate, train, verify, and operate that system.

Setup and configuration

The package can be run directly with npx. The documented examples use version 0.9.1:

• npx @ruvnet/ruview@0.9.1 doctor checks the local setup and provides source-cited guidance.
• npx @ruvnet/ruview@0.9.1 guidance --topic sensing --query "model loading" requests guidance for a specific topic.
• npx @ruvnet/ruview@0.9.1 devices reports connected devices.
• npx @ruvnet/ruview@0.9.1 esp32 --watch shows a live view for ESP32 and Realtek RAC1 nodes.
• npx @ruvnet/ruview@0.9.1 esp32 --seconds 45 --analyze analyzes live CSI through the vitals kernel.

The MCP server is documented as usable over stdio or HTTP. The supplied material does not specify required environment variables, authentication settings, or a complete client configuration file.

Tools and capabilities

Depending on the selected workflow, @ruvnet/ruview can provide:

• Source-cited operational guidance.
• Device discovery and host diagnostics.
• Access to CSI nodes, radar, and LiDAR.
• Live ESP32 CSI viewing and analysis.
• Verified ESP32 flashing with boot evidence.
• Calibration and evidence-gated model training.
• Deterministic verification and checks on sensing claims.
• SSH access to remote hosts.
• Cognitum Spaces support.

Limitations and notes

The supplied documentation does not list the MCP tool names, HTTP route, request schemas, authentication model, or supported operating systems. It also does not establish that every sensing capability is production-validated: the README explicitly notes that the unified RF world model remains synthetic until real-data validation. Treat inferred vitals, pose, and activity results as dependent on the hardware, calibration, models, and evidence available in the deployment.