Hardware Probe

io.github.yamaru-euPublished by GitHub account yamaru-euv1.1.2Updated Oct 24, 2026

Hardware probe for performance diagnostics, thermal monitoring, and LLM optimization.

VerifiedSTDIODesktop onlySecurity & MonitoringAI & ML

Overview

AI-generated overview

Reads local hardware, performance, thermal, and storage data so an assistant can diagnose system slowness and plan local LLM runs.

What you can do
A local MCP server that probes the machine it runs on. Tools cover full hardware inventory (CPU, RAM, GPU, storage, OS), real-time performance and top processes, RAM pressure, disk SMART health, thermal and throttling states, antivirus/EDR impact, and statistical health reports over a time window. Two beta tools predict LLM performance and recommend local models through a remote API, and another inspects AI runtime configuration.
When to use it
Useful when troubleshooting a slow or throttling machine, checking whether a local model will fit in available VRAM, or tuning local inference runtimes such as Ollama, CUDA, or Metal. Best for desktop users who want the assistant to reason about their actual hardware rather than generic specs.
Requirements
Runs locally over stdio, installed via npm as @yamaru-eu/hardware-probe (Node.js and npx required). Desktop only; no web execution. The optional remote recommendation tools need the REMOTE_API_URL environment variable pointing at the remote API. No authentication is declared.
Before you install
The server reads detailed system information, including process lists, hardware identifiers, and AI runtime environment variables. The README states unique hardware identifiers are anonymized before remote transmission, but the beta LLM tools do send data to the endpoint set in REMOTE_API_URL, so review what leaves the machine. All tools appear read-only; no writes or deletions are described.

Installation

In SourceWeft

  1. Open Hardware Probe in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Yamaru Hardware Probe (MCP)

[NPM Version] [NPM Downloads] [MCP Registry] [Awesome MCP] [PRs Welcome] [License]

Expert system hardware probe and performance diagnostic engine for AI, Gaming, and High-Performance workflows. This is a Model Context Protocol (MCP) server that provides deep system insights beyond simple specifications.

Key Features

  • 🔍 Deep Hardware Inventory: Comprehensive analysis of CPU, RAM, GPU (VRAM/Bandwidth), Storage, and OS topology.
  • ⚡ Real-time Performance Monitoring: Live tracking of system load and identification of resource-hogging processes.
  • 🧊 Thermal & Power Diagnostics: Detects thermal throttling and frequency clipping to resolve unexpected slowness.
  • 🤖 AI/LLM Optimization: Specialized tools for predicting LLM performance, calculating quantization fit, and optimizing runtimes (Ollama, CUDA, Metal).
  • 🛡️ Privacy-First: Automatic anonymization of unique hardware identifiers before any remote transmission.

Installation

For Gemini CLI Users

bash
gemini extension install @yamaru-eu/hardware-probe

For Manual MCP Setup

Add this to your MCP settings file (e.g., npx-config.json or claude_desktop_config.json):

json
{  "mcpServers": {    "yamaru-probe": {      "command": "npx",      "args": ["-y", "@yamaru-eu/hardware-probe"]    }  }}

Available Tools

  • analyze_local_system: Full hardware inventory.
  • analyze_performance: Real-time performance metrics and top processes.
  • analyze_ram_pressure: Detailed memory pressure and RSS analysis for deep RAM troubleshooting.
  • check_storage_health: Disk SMART health, firmware, and I/O bottleneck analysis.
  • thermal_profile: Real-time CPU/GPU thermal states, fan speeds, and frequency throttling detection.
  • diagnose_antivirus_impact: Detects EDR/Antivirus conflicts and exclusion coverage on dev paths.
  • monitor_system_health: Statistical health report (CPU load, RAM usage, temperature) with min/max/avg over a configurable time window (up to 10 minutes).
  • check_llm_compatibility (BETA): Predicts performance for a specific LLM model via remote API.
  • get_llm_recommendations (BETA): Recommends the best local models via remote API.
  • analyze_inference_config: Deep-dive into AI runtimes and environment variables.

Skills Integration

When used with Gemini CLI, this extension provides the following expert skills:

  • hardware-performance-expert: Global protocol for system health and troubleshooting.
  • local-inference-optimizer: Specialized logic for fine-tuning local LLM runs.

Development

bash
npm install        # Install dependenciesnpm run build      # Compile TypeScript → dist/npm run test       # Run test suitenpm run inspector  # Test tools in the MCP Inspector

License

Apache 2.0 - Part of the Yamaru Project.

Source: README.md at commit fa22f84

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v1.1.2LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts