linux-mcp is a Linux system monitoring MCP server built on the Model Context Protocol (MCP), providing real-time system information over STDIO transport. The server is designed to be read-only and does not modify system state.
Purpose linux-mcp is intended to enable MCP clients (such as Claude Desktop, Claude Code, Codex, OpenCode, or any MCP host) to query the real-time state of a Linux system. It exposes system data through MCP tools and resources, making it suitable for scenarios that need to integrate system monitoring capabilities into AI assistants or automated workflows.
Main capabilities The server provides 65 tools and 19 resources, covering the following categories:
- System: hostname, operating system, kernel, architecture, uptime, load average
- CPU: usage, model, frequency, core count, temperature sensors
- Memory: RAM and swap usage and percentages
- Disk: usage per partition, inodes, mount options, max files, block devices
- Network: interface statistics, active connections, listening ports, DNS, ping, IP geolocation and ASN lookup
- Processes: running processes sorted by CPU or memory, open file descriptors per process
- Docker: containers, images, networks, volumes, disk usage, system information, and container statistics
- Services and automation: systemd units, service status, user timers, crontab, system cron jobs
- Security: active user sessions, failed login detection, SELinux/AppArmor status, firewall/SSH/SUID/world-writable audits, and security scoring
- Packages: installed packages and available updates (pacman, dpkg, rpm/dnf, apk)
- Hardware: GPU information, PCI/USB bus devices, power/battery analysis
- Desktop session: Wayland/X11 protocol, desktop environment identification, runtime configuration
- Storage health: RAID status, logrotate configuration, time synchronization, SMART disk health, per-device I/O metrics
- System health: comprehensive health assessment, including memory, disk, load, and systemd checks
- /proc diagnostics: in-depth /proc inspection, including interrupts, softirqs, vmstat, diskstats, filesystems, kernel version, slabinfo
- man pages: system manual pages for any installed command
- snapshots: get a comprehensive system overview in a single call
- MCP resources: system data is also accessible as readable resources
Usage The server communicates over STDIO transport, follows the MCP standard, and is designed to be launched by an MCP client.
Installation options:
- Via MCP Registry: discover the server on the MCP Registry and install it through an MCP client (one-click install in VS Code or manual configuration).
- Download precompiled binaries: download the latest binary from the GitHub Releases page. A fully static build (no libc dependency), linux-mcp_static, is also available. Place the binary in $GOPATH/bin/ (usually ~/go/bin/).
- Build from source: requires Go 1.27+. The binary is placed in bin/linux-mcp. A fully static binary (no libc dependency) can also be built.
Requirements:
- Linux (the server targets Linux; some tools use Linux-specific paths)
- Go 1.27+ (only required when building from source)
- Docker (optional, only required for Docker tools)
Direct run: starts the server and listens for MCP requests on STDIN/STDOUT.
Client configuration examples:
- OpenCode: add the server configuration to the OpenCode config, or add it via the CLI. It can be scoped to all projects rather than just the current project.
- Claude Code: create a .mcp.json file in the project root for team-shared configuration, or verify the connection with /mcp in a Claude Code session.
- Codex: manually add it to ~/.codex/config.toml (or a project-scoped .codex/config.toml), and verify the connection with codex mcp list.
Configuration: The server can be configured via a JSON file loaded from ~/.config/linux-mcp/config.json or the LINUX_MCP_CONFIG environment variable. The server also handles SIGHUP to reload the configuration file at runtime without restarting.
Use cases
- Linux system monitoringintegrated into AI assistants or MCP clients
- Query system status, processes, network, Docker containers, and other information via natural language
- Security auditing and system health checks
- Quickly obtain system snapshots in development or operations scenarios
Limitations
- The server is designed to be read-only and does not modify system state
- It targets Linux systems, and some tools depend on Linux-specific paths
- Docker tools require a Docker environment
- Building from source requires Go 1.27+
- The server is published on the MCP Registry under the identifier io.github.Mohabdo21/linux-mcp
Dependencies and acknowledgements
- modelcontextprotocol/go-sdk - MCP SDK for Go
- shirou/gopsutil/v4 - system metrics (CPU, memory, disk, network, processes, sensors)
- docker/go-sdk - Docker Engine API client
- moby/moby/client - Docker client library
- ip-api.com - free IP geolocation API (used by get_ip_info)
