rhel-lightspeed/linux-mcp-server

Tools to allow LLM clients to interact with Linux systems remotely

What it solves

This project provides a safe, read-only interface for AI models to perform Linux system administration, diagnostics, and troubleshooting, specifically optimized for RHEL-based systems. It prevents the AI from making destructive changes to the system while allowing it to gather the necessary information to diagnose issues.

How it works

It implements the Model Context Protocol (MCP) to act as a bridge between an LLM and a Linux environment. The server can execute read-only commands on local or remote hosts via SSH with key-based authentication, allowing an AI agent to query system information, services, processes, logs, and network status.

Who it’s for

System administrators and DevOps engineers working with Red Hat Enterprise Linux (RHEL) who want to use AI agents to automate the diagnostic phase of troubleshooting.

Highlights

  • Read-Only Safety: All tools are strictly read-only to ensure system stability.
  • Remote Execution: Supports SSH for managing multiple remote hosts within a single session.
  • RHEL Optimized: Specifically designed for systemd and RHEL-based environments.
  • Controlled Log Access: Access to log files can be restricted via environment variables.

Related

  • Project
    neon-solutions/add-mcp

    A CLI tool that allows users to to install and manage Model Context Protocol (MCP) servers across multiple AI coding agents with a single command.

  • Project
    ctrlb-hq/ctrlb-decompose

    A log analysis tool that collapses millions of noisy log lines into a few typed patterns, providing LLM-optimized output and local-first processing.

  • Project
    Gan-Tu/Widgets

    A compact, schema-capable widget renderer for chat UIs that allows AI agents to generate interactive UI components using a strict template language and a large library of pre-built components.

  • Project
    NX-AI/tirex

    TiRex is a pre-trained xLSTM-based model for zero-shot time series forecasting, classification, and regression, providing high-performance predictions across both long and short horizons without requiring training on user data.