M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
C

CLIO Node Hardware

Community
Comprehensive hardware monitoring and system analysis
How to install ↓ Compare GitHub source repository ↗
Category
Dev Tools #158 of 438
Stars
★ 30 Popular
Transport
stdio (local process)
Runtime
Python 3.10+
Credentials
No credential needed
License
BSD-3-Clause
Last commit
Tools
11
57FMRS · C

CLIO Node Hardware is part of CLIO Kit and provides comprehensive hardware monitoring tools, suitable for AI agents in HPC environments for system management and troubleshooting. It provides reliable information via stdio.

Strongest · Maintenance 13/20 Weakest · Reliability 9/20

Reliability
9/20
Security and permissions
12/20
Maintenance
13/20
Documentation
11/20
Setup experience
12/20
Why each score
Reliability 9/20
The manifest declares 11 tools, one resource, and one prompt with clear, well-scoped descriptions; the static structure needed for the init and tool-list handshake is complete. However, no server-specific source, committed tests, or CI evidence for node-hardware is supplied (repo-wide CI/coverage badges cannot be attributed to this server), and paths depending on external tools (SSH, nvidia-smi/rocm-smi) lack documented error handling or degradation, so reliability cannot exceed 12.
Security and permissions 12/20
get_remote_node_info collects data from remote nodes via SSH — a clear data-boundary expansion with no disclosure of credential handling (agent? keys?), host allowlisting, or command-injection protections; this drives the deduction. Remaining tools are read-only monitoring; no delete/pay/remote-execution defaults, malware, or credential theft observed, so it is not a red-line case.
Maintenance 13/20
The repo is not archived, has 39 open issues indicating real usage and a response channel, a clear BSD-3-Clause license, named ownership (GRC at IIT, NSF funding). However, versioning is inconsistent: manifest server version 2.2.3 vs PyPI package 2.10.6 vs README install example 2.4.3, indicating governance gaps.
Documentation 11/20
The README provides solid repo-level installation, client configs (Cursor/Claude Code/VS Code/Claude Desktop), and troubleshooting, and the server table lists node-hardware 2.2.3 with its install command. But there is no server-specific documentation of tool parameters, SSH/GPU-tool prerequisites, or limitations (sensor availability in containers, platform coverage); repo-level docs are substituted for server-level detail.
Setup experience 12/20
The setup path is clear: uv tool install then 'clio-kit mcp-server node-hardware', with multi-client JSON config examples that can be adapted. Deductions: no ready-made client snippet for node-hardware itself, and the stdio path depends on uv installation and PATH setup, adding manual steps.

Static review · not runListed 2026-09-08

Read the FMRS scoring method →

Fit and risk

What it can accessUses the network

Best for

  • AI agents that need to access hardware information in natural language
  • Core developers who need quick insight into system resources
  • Users managing scientific HPC workloads

Not for

  • Detailed benchmarking or low-level hardware raw data (beyond this server's scope)
  • Performance monitoring on remote machines where SSH is not possible
  • Monitoring systems that require advanced aggregation across multiple nodes (only single remote node fetching is provided here)

Required permissions

  • Read CPU, memory, disk, network, GPU, sensor, and process information
  • Execute read-only commands locally (such as nvidia-smi/rocm-smi)
  • SSH access to collect remote node information

Risks and side effects

  • Process and system information may contain sensitive data if exposed to the AI agent
  • GPU queries may fail due to driver or hardware issues
  • Remote monitoring requires valid SSH credentials and may pose security risks

Setup

Before you start

Runtime：Python 3.10+

  1. Install uv package manager: curl -LsSf https://astral.sh/uv/install.sh | sh or pip install uv.
  2. Install clio-kit: uv tool install 'clio-kit==2.4.3' and run uv tool update-shell.
  3. Configure the MCP server in your client, for example: {"mcpServers":{"node-hardware":{"command":"clio-kit","args":["mcp-server","node-hardware"]}}}
claude_desktop_config.json 
{
  "mcpServers": {
    "node-hardware": {
      "command": "clio-kit",
      "args": [
        "mcp-server",
        "node-hardware"
      ]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "node-hardware": {
      "command": "clio-kit",
      "args": [
        "mcp-server",
        "node-hardware"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add node-hardware -- clio-kit mcp-server node-hardware

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Start the server with `clio-kit mcp-server node-hardware` and confirm tools like get_cpu_info and health_check appear in your MCP client's tool list, then call health_check to verify monitoring capability status.

Troubleshooting

  1. Ensure clio-kit is installed via uv tool install and update-shell has been run
  2. Check MCP client configuration for correct command and arguments
  3. Verify that GPU tools (nvidia-smi or rocm-smi) are available and executable on the system
  4. For remote nodes, check SSH connectivity, credentials, and firewall settings

Things to try

Once connected, you can ask your AI assistant things like:

  • Get CPU specifications and per-core usage on this node.
  • Check memory and swap usage percentages.
  • Get GPU model, temperature, and utilization.
  • Run a full system health check including disk and network stats.

Tools 11

get_cpu_info read-only
Get CPU specifications, core counts, frequencies, and per-core usage.
get_memory_info read-only
Get RAM and swap capacity, usage percentages, and availability.
get_system_info read-only
Get OS details, hostname, uptime, and active users.
get_disk_info read-only
Get disk partitions, usage statistics, and I/O counters.
get_network_info read-only
Get network interfaces, IP addresses, and I/O statistics.
get_gpu_info read-only
Get GPU model, memory, temperature, and utilization via nvidia-smi/rocm-smi.
get_sensor_info read-only
Get temperature, fan speed, and battery sensor readings.
get_process_info read-only
Get running processes with CPU, memory, and status details.
Show 3 more tools
get_performance_info read-only
Get real-time CPU, memory, disk, and network performance metrics.
get_remote_node_info read-only
Collect hardware info from a remote node via SSH. Supports component filtering.
health_check read-only
Verify server health and hardware monitoring capability status.

Use cases

Get hardware specs and status of the current node
Monitor hardware health of remote nodes
Collect performance metrics for system tuning

Supported clients

Claude Desktop
Cursor
VS Code

Listed from the project's documentation, not tested by this site.

Overview

CLIO Node Hardware (originally io.github.iowarp/node-hardware-mcp) is an MCP server that provides comprehensive hardware monitoring and system analysis capabilities for LLMs. It runs over stdio and offers detailed information on CPU, memory, disk, network, GPU, sensors, processes, and real-time performance, with support for remote node monitoring via SSH.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →
Compare with other MCP servers

More servers from this repository

CLIO Arxiv MCP ServerCLIO CompressionHDF5 MCP ServerCLIO LmodCLIO PandasCLIO ParaViewCLIO ParquetCLIO Plot MCP ServerCLIO SeismicCLIO SlurmCLIO SpackCLIO Web MCP Server

Source revision 93e41d70d6c2 Data synced 2026-10-19 Read the FMRS scoring method