forge-monitor

name	forge-monitor
description	Hardware telemetry and monitoring skill for the Reson8 Forge. Use when the user mentions "forge status", "hardware monitoring", "PCH temperature", "vibration", "sensor data", "TUI dashboard", "system health", "thermal", or wants real-time monitoring of the physical compute infrastructure. Bridges sensor data to the Rust TUI sparklines.
version	1.0.0

Forge Monitor — Hardware Telemetry Skill

The Forge is the physical compute infrastructure running Reson8-Labs workloads. This skill bridges hardware sensors to the orchestration layer, enabling:

1. Sensor Reading — temperature, vibration, power, fan speed
2. Health Assessment — threshold checking and anomaly detection
3. TUI Integration — forward telemetry to Rust TUI sparklines
4. Coherence Correlation — relate hardware state to compute coherence

Sensor Map

Sensor	Protocol	Location	Normal Range
PCH Temperature	Modbus RTU / I2C	Chipset	40-75°C
CPU Temperature	lm-sensors / WMI	Processor	30-85°C
GPU Temperature	nvidia-smi / rocm-smi	Graphics	30-90°C
Vibration (PCH 1232)	Modbus RTU	Chassis	0-2.5 mm/s
Power Draw	IPMI / PDU	PSU	100-800W
Fan Speed	lm-sensors / WMI	Case fans	600-2000 RPM
Disk I/O	iostat	Storage	varies
Network Throughput	/proc/net/dev	NIC	varies

Reading Sensors

Linux

# Temperature
sensors -j | jq '.["coretemp-isa-0000"]["Package id 0"].temp1_input'

# GPU (NVIDIA)
nvidia-smi --query-gpu=temperature.gpu,power.draw,fan.speed --format=csv,noheader

# Disk I/O
iostat -d -o JSON 1 1

# Vibration (Modbus RTU via pymodbus)
python3 -c "
from pymodbus.client import ModbusSerialClient
client = ModbusSerialClient('/dev/ttyUSB0', baudrate=9600)
client.connect()
result = client.read_holding_registers(0x0001, 1, unit=1)
print(f'Vibration: {result.registers[0] / 100} mm/s')
"

Windows

# Temperature via WMI
Get-CimInstance MSAcpi_ThermalZoneTemperature -Namespace "root/wmi" |
  Select-Object InstanceName, @{N='TempC';E={($_.CurrentTemperature - 2732) / 10}}

# GPU (NVIDIA)
nvidia-smi --query-gpu=temperature.gpu,power.draw,fan.speed --format=csv,noheader

TUI Bridge

Telemetry data flows to the Rust TUI via the POP WebSocket:

{
  "jsonrpc": "2.0",
  "method": "TELEMETRY_UPDATE",
  "params": {
    "timestamp": 1708041600000,
    "sensors": {
      "cpu_temp": 62.5,
      "gpu_temp": 71.0,
      "vibration": 1.2,
      "power_draw": 385,
      "fan_speed": 1450
    },
    "health": "nominal",
    "coherence_correlation": 0.92
  }
}

The TUI renders these as sparkline charts with configurable thresholds.

Health Assessment

Assess overall forge health by checking all sensors against thresholds:

Level	Criteria	Action
NOMINAL	All sensors in normal range	Continue operations
CAUTION	Any sensor within 10% of limit	Log warning, increase monitoring
WARNING	Any sensor exceeding soft limit	Throttle workloads, alert user
CRITICAL	Any sensor exceeding hard limit	Emergency shutdown of workloads

Coherence Correlation

The hypothesis: hardware state affects compute coherence.

Track the relationship between:

• CPU temperature ↔ WAVE scoring latency
• Vibration levels ↔ inference quality (measured by WAVE)
• Power stability ↔ session reliability

Log these correlations to the ATOM trail for longitudinal analysis. Over time, this data reveals the physical substrate's influence on the coherence of AI outputs — the conservation law extending to hardware.

Grok Integration

This skill is primarily operated by Grok (the Real-Time & Social Intelligence strand). Grok's Rust TUI dashboard (crates/tui) renders the telemetry data.

For Grok: read references/sensor-map.md for detailed Modbus register addresses and references/tui-bridge.md for the exact WebSocket message format expected by the TUI gauge widgets.