linux-mcp
Linux system monitoring and diagnostics MCP server over STDIO, exposing many mostly read-only tools and resources.
Collect host, CPU, memory, load, temperature, uptime, hardware/BIOS/TPM, desktop-session, power, GPU, PCI/USB, kernel-module, and /proc diagnostics.
Inspect disk usage, inodes, mounts, block devices, largest files, disk I/O, RAID, SMART health, and logrotate/time-sync status.
Monitor network interfaces, connections/listening sockets, routing, DNS, ping, and IP geolocation/ASN.
Analyze processes: top CPU/memory/I/O, process tree, open file descriptors, file locks, shared memory, and user sessions.
Work with Docker: list/inspect containers, images, networks, volumes, logs, stats, diffs, disk usage, and daemon info.
Review systemd services/units/timers, boot timing/blame/critical chain, cron/user automation, journal/audit logs, failed logins, and man pages.
Run security/health checks and inventory packages/updates, users/groups, SELinux/AppArmor, firewall/SSH/SUID/world-writable files, environment variables, and system snapshots.
Configure timeouts, disabled tools, log level, and reload configuration via SIGHUP.
Provides tools for monitoring Docker containers, images, networks, volumes, disk usage, system info, and stats, enabling management and inspection of Docker resources on the host.
Provides tools for detecting the Wayland display protocol and desktop environment identifiers, as well as runtime configuration details for the desktop session.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@linux-mcp check system health"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
linux-mcp - Linux MCP Server
A Linux system monitoring server built on the Model Context Protocol (MCP). Provides real-time system information - CPU, memory, disk, network, processes, Docker, and more - via MCP tools over STDIO transport.
Features
System - hostname, OS, kernel, architecture, uptime, load averages
CPU - usage, model, frequency, core counts, temperature sensors
Memory - RAM and swap usage with percentages
Disk - per-partition usage, inodes, mount options, largest files, block devices
Network - interface stats, active connections, listening ports, DNS, ping, IP geolocation & ASN lookup
Processes - running processes sorted by CPU or memory, open file descriptors per process
Docker - containers, images, networks, volumes, disk usage, system info, and container stats
Services & 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 audit with security score
Packages - installed packages and available updates (pacman, dpkg, rpm/dnf, apk)
Hardware - GPU info, PCI/USB bus devices, power/battery analytics
Desktop session - Wayland/X11 protocol, DE identifiers, runtime config
Storage health - RAID status, logrotate configuration, time synchronization, SMART disk health, per-device I/O metrics
System health - comprehensive health assessment with memory, disk, load, and systemd checks
/proc diagnostics - deep /proc inspection: interrupts, softirqs, vmstat, diskstats, filesystems, kernel version, slabinfo
Man pages - system manual pages for any installed command
Snapshot - comprehensive system overview in a single call
MCP Resources - system data also accessible as readable resources
Related MCP server: DivLens MCP
Prerequisites
Go 1.27+ (to build from source)
Linux (the server targets Linux; some tools use Linux-specific paths)
Docker (optional - only needed for Docker tools)
Installation
Via MCP Registry
Discover the server on the MCP Registry and install via your MCP client (VS Code one-click install, or manual config).
Download pre-built binary
Download the latest binary from the GitHub Releases page:
curl -LO https://github.com/Mohabdo21/linux-mcp/releases/latest/download/linux-mcp
chmod +x linux-mcpA fully static build (no libc dependency) is also available as linux-mcp_static.
Via go install
go install github.com/Mohabdo21/linux-mcp@latestThe binary is placed in $GOPATH/bin/ (typically ~/go/bin/).
Build from source
git clone https://github.com/Mohabdo21/linux-mcp.git
cd linux-mcp
make buildThe binary is placed at bin/linux-mcp.
For a fully static binary (no libc dependency):
make build-staticUsage
The server communicates over STDIO transport, following the MCP standard. It is designed to be launched by an MCP client (e.g., Claude Desktop, Claude Code, Codex, OpenCode, or any MCP host).
Running directly
./bin/linux-mcpThis starts the server and listens for MCP requests on STDIN/STDOUT.
Integration
Add the following to your OpenCode configuration:
{
"mcp": {
"linux-mcp": {
"type": "local",
"command": ["/path/to/linux-mcp"],
"enabled": true
}
}
}Add the server via the CLI:
claude mcp add linux-mcp -- /path/to/linux-mcpTo scope it to all your projects instead of just the current one:
claude mcp add linux-mcp --scope user -- /path/to/linux-mcpOr create a .mcp.json file in your project root for team-shared configuration:
{
"mcpServers": {
"linux-mcp": {
"type": "stdio",
"command": "/path/to/linux-mcp"
}
}
}Verify the connection inside a Claude Code session with /mcp.
Add the server via the CLI:
codex mcp add linux-mcp -- /path/to/linux-mcpOr add it manually to ~/.codex/config.toml (or .codex/config.toml for project-scoped config):
[mcp_servers.linux-mcp]
command = "/path/to/linux-mcp"Verify the connection with codex mcp list.
Tools and resources
65 tools and 19 resources covering system, CPU, memory, disk, network, processes, Docker, security, packages, hardware, and more.
Full tool and resource reference
Configuration
The server can be configured via a JSON file loaded from ~/.config/linux-mcp/config.json or the LINUX_MCP_CONFIG environment variable.
{
"log_level": "info",
"timeouts": {
"get_system_snapshot": "60s",
"ping_host": "5s"
},
"disabled": []
}Field | Description |
| One of |
| Per-tool timeout overrides as Go duration strings |
| Tool names to disable, including their resource and aggregate collector paths |
Invalid configuration prevents startup. The server also handles SIGHUP to reload the configuration file at runtime without restarting; a failed reload retains the last valid configuration. Disabled collectors are skipped by aggregate tools and reported in their errors.
Docker diagnostic responses omit environment values and raw command arguments, including image-history commands. Other diagnostic content, such as logs and labels, can still contain sensitive data; grant access only to trusted clients.
MCP Registry
Published on the MCP Registry as io.github.Mohabdo21/linux-mcp.
License
Dependencies
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)
Available Tools
65 toolsget_audit_logsA
Returns Linux audit (auditd) events from /var/log/audit/audit.log: AVC denials and syscall records. lines defaults to 50. Read-only; fatal if the file is unreadable (auditd not installed or not readable). Use for auditd denial and syscall forensics; get_journal_logs for the systemd journal and kernel messages, get_failed_logins for login attempts.
Inputs: lines. Returns: errors, entries
get_available_updatesA
Lists available package updates without applying them (pacman -Qu, apt list --upgradable, dnf check-update, apk version -l <). Read-only, no cache refresh, nothing is installed. A clean manager with nothing to report returns an empty list. Fatal only if no supported package manager is present. Use to see pending updates; get_installed_packages for current versions.
Returns: total, errors, updates
get_block_devicesA
Returns block devices and partitions: names, sizes, filesystem types, and mount points. Read-only from sysfs and /proc/mounts. Loop, ram, and zram devices are skipped. Errors land in the errors field. Use for storage inventory; get_disk_info for usage percent.
Returns: errors, devices
get_boot_blameA
Returns systemd units ranked by their own start duration from systemd-analyze blame. Read-only; fatal only if systemd-analyze is missing or fails. Use for a flat ranking of the slowest units by self-time; get_boot_critical_chain for the dependency path that gates the target, get_boot_time for phase totals.
Returns: errors, entries
get_boot_critical_chainA
Returns the time-critical boot chain from systemd-analyze critical-chain: a dependency tree of units with their active-time point and start duration. Read-only; fatal only if systemd-analyze is missing or fails. When unit is omitted, returns the full chain to the default target (usually graphical.target). When unit is provided (e.g. 'nginx.service'), returns only the chain from that unit downward. Use to find the dependency path that delays the boot target; get_boot_blame for a flat per-unit ranking, get_boot_time for phase totals.
Inputs: unit. Returns: chain, errors, target
get_boot_timeA
Returns the phase-level boot breakdown from systemd-analyze time: firmware, loader, kernel, initrd, and userspace durations plus total startup and the reached target. Read-only; fatal only if systemd-analyze is missing or fails. Use to see which boot phase dominates; get_boot_blame ranks slow units, get_boot_critical_chain traces the dependency path.
Returns: total, errors, phases, target, +3 more
get_cpu_infoA
Returns CPU model, frequency, core count, and current usage percent. Read-only from /proc/stat and /proc/cpuinfo. Fatal only if either source is unreadable. Use for capacity planning or busy/idle checks.
Returns: cores, errors, usage_percent, physical_core_count
get_cpu_temperatureA
Returns current CPU temperature from hwmon sensors if available. Read-only. If no sensors are exposed, returns a message in the errors field instead of failing. Use for thermal monitoring; prefer get_system_health_check for an overall verdict.
Returns: errors, message, temperatures
get_cron_jobsA
Returns system-level cron jobs from /etc/crontab and the periodic cron directories (/etc/cron.daily, .weekly, .hourly). Read-only; no root needed. Missing paths land in the errors field. Use for system automation; get_user_automation for user-level tasks.
Returns: errors, daily_jobs, hourly_jobs, weekly_jobs, +1 more
get_desktop_session_infoA
Returns display protocol (Wayland/X11), desktop environment, and related environment config. Read-only from environment variables; never fails. Use to understand the GUI session the server runs under.
Returns: errors, display, runtime_dir, session_type, +2 more
get_disk_infoA
Returns disk usage per mounted partition. Params: mount_point (exact match, must start with /) and threshold (only partitions at or above it). Read-only via statfs. Fatal only if partitions cannot be listed; per-partition read errors are skipped. Use for capacity; get_inode_usage when 'disk full' persists with free space.
Inputs: threshold, mount_point. Returns: errors, partitions
get_disk_io_metricsA
Returns per-device disk I/O counters cumulative since boot: reads, writes, sectors, and timings. Read-only from /proc/diskstats (loop/ram/zram skipped); fatal only if unreadable. Use to attribute I/O across devices; get_top_io_processes for the processes driving it.
Returns: errors, metrics
get_docker_container_detailsA
Returns a container's state, config, environment names, mounts, and network settings. Environment values and raw command arguments are omitted. Read-only via the Docker API; fatal if the daemon is unreachable or the container is unknown. Use to inspect one container; get_docker_info to list them.
Inputs: container_id (required). Returns: errors, container
get_docker_container_diffA
Returns filesystem changes (added, modified, deleted) in a container since it started. Read-only via the Docker API. Use to see what a container wrote to its writable layer.
Inputs: container_id (required). Returns: errors, changes
get_docker_container_logsA
Returns a container's stdout/stderr log lines. Params: tail (line count) and timestamps. Read-only via the Docker API. Fatal if the daemon is unreachable or the container is unknown. Log content may contain sensitive data.
Inputs: container_id (required), tail, timestamps. Returns: logs, errors
get_docker_container_statsA
Returns live CPU, memory, network I/O, and PID stats for running containers. Params: container_ids comma-separated or 'all' (running only). Read-only, one-shot sample. Stats only exist for running containers; unknown or stopped ids error. Use container_ids 'all' to cover every running container in one call.
Inputs: container_ids (required). Returns: errors, containers
get_docker_container_topA
Returns container processes with fixed PID, PPID, user, state, and executable-name columns, without command arguments. Custom args are ignored. Read-only via the Docker API; fatal if the daemon is unreachable or the container is unknown. Use to debug container-level process state.
Inputs: container_id (required), args. Returns: errors, titles, processes
get_docker_disk_usageA
Returns Docker disk usage broken down by containers, images, volumes, and build cache. Read-only via the Docker API. Use to find what is consuming disk space.
Returns: errors, images, volumes, containers, +1 more
get_docker_image_detailsA
Returns an image's config, environment names, labels, and layers. Environment values and raw command arguments are omitted. Read-only via the Docker API; fatal if the daemon is unreachable or the image is unknown. Use to inspect a single image; get_docker_image_history for layer metadata.
Inputs: image_id (required). Returns: image, errors
get_docker_image_historyA
Returns an image's layer history: sizes and creation times. Raw build commands are omitted. Read-only via the Docker API. Use to understand what an image is built from; get_docker_image_details for config and labels.
Inputs: image_id (required). Returns: errors, layers
get_docker_infoA
Returns Docker containers (including stopped) and images. Read-only via the Docker API. Fatal if the daemon is unreachable. Use for the Docker inventory of containers and images; get_docker_container_details to inspect one container, get_docker_disk_usage for space accounting.
Returns: errors, images, containers
get_docker_networksA
Returns Docker networks with driver, scope, and configuration details. Read-only via the Docker API. Use for network topology debugging.
Returns: errors, networks
get_docker_system_infoA
Returns Docker daemon info: version, storage driver, runtimes, and resource counts. Read-only via the Docker API; fatal if the daemon is unreachable. Use to check daemon config and health.
Returns: info, errors
get_docker_volumesA
Returns Docker volumes with driver, mountpoint, size, and labels. Read-only via the Docker API. Use to inventory storage; get_docker_disk_usage for space accounting.
Returns: errors, volumes
get_environment_variablesA
Returns the server process's environment as a sorted key-value map. search filters by name prefix or substring, case-insensitive. Read-only (os.Environ). Sensitive names (SECRET, TOKEN, PASSWORD, etc.) are redacted to ***. Use to debug PATH, locale, and server configuration.
Inputs: search. Returns: count, errors, variables
get_failed_loginsA
Returns recent failed login attempts (excluding Boot records) with summary statistics. Read-only: lastb -n N, falling back to journalctl if btmp is not readable. Default 20 entries. Errors are non-fatal. Use for login-attempt triage; get_audit_logs for auditd AVC and syscall events.
Inputs: lines. Returns: errors, entries, summary
get_file_locksA
Returns active file locks from /proc/locks: type, mode, PID, byte range, and path. Read-only; an unreadable file is silently swallowed, parse errors go in the errors field. Use to find who holds a lock.
Returns: locks, errors
get_gpu_infoA
Returns GPU usage, memory, temperature, and power draw. Read-only, via nvidia-smi, rocm-smi, or intel_gpu_top in that order of availability. Fatal only if no GPU tool is installed. The Intel fallback reports presence, not metrics. Use for GPU workload monitoring.
Returns: gpus, errors, vendor
get_hardware_bus_infoA
Lists PCI and USB devices detected on the system. Read-only via lspci/lsusb with a sysfs fallback; if both fail the error goes in the errors field. search filters any field (bus, slot, class, vendor, device). Use to identify network cards, audio interfaces, and expansion cards for driver troubleshooting.
Inputs: search. Returns: errors, pci_devices, usb_devices
get_inode_usageA
Returns inode usage per filesystem via df -i. mount_point is an optional exact filter. Read-only; fatal if df is missing or fails. Use when 'disk full' errors persist despite free space - inode exhaustion.
Inputs: mount_point. Returns: errors, mounts
get_installed_packagesA
Queries installed packages: pacman -Q on Arch, dpkg -l on Debian, rpm -qa on rpm/dnf, apk list --installed on Alpine. Optional name filter, resolved by the package manager itself. Read-only. Fatal only if no supported package manager is present or the query fails. Use to check what is installed; get_available_updates for pending upgrades.
Inputs: name. Returns: total, errors, packages
get_ip_infoA
Returns geolocation, ASN/organization, and provider tags (e.g. AWS, Cloudflare) for an IP or your public IP. Makes an external call to ip-api.com - network-dependent and can be slow offline. Invalid IPs return an error in the errors field. Use for network egress and remote-peer context.
Inputs: ip. Returns: ip, asn, org, city, +4 more
get_journal_logsA
Reads systemd journal entries with optional filtering by unit, priority, and time range (since/until). Set user=true for the user journal. Read-only via journalctl. Fatal only if journalctl fails or an invalid unit is given. Returns structured entries with timestamp, message, priority, unit, and PID. For auditd AVC/syscall events use get_audit_logs.
Inputs: unit, user, lines, since, +2 more. Returns: errors, entries
get_kernel_modulesA
Returns loaded kernel modules with size, reference count, and dependency info. Read-only from /proc/modules. Use for driver and module troubleshooting.
Returns: total, errors, modules
get_largest_filesA
Lists the top N largest entries in a directory by size (default path '.', limit 10-100). Read-only and non-recursive - only one directory level, and directory sizes show their own stat size, not contents. Failures silently return empty results. Use for quick space triage; get_disk_info for partition-level capacity.
Inputs: path, limit. Returns: path, errors, entries
get_load_averageA
Returns 1-, 5-, and 15-minute load averages. Read-only from /proc/loadavg; fatal only if unreadable. Use as a quick utilization check; get_system_health_check compares load against core count.
Returns: errors, load_1, load_5, load_15
get_logged_in_usersA
Returns active user sessions via who -u: username, terminal, origin, and login time. Read-only; fatal only if who is missing. Use for security awareness and multi-user workload checks.
Returns: users, errors
get_logrotate_statusA
Returns logrotate configuration files and the state file path. Read-only from /etc/logrotate.conf and /etc/logrotate.d. Errors land in the errors field. Use to verify rotation policies.
Returns: errors, configs, state_file
get_man_pageA
Fetches the authoritative man page for a command as plain text. command is required and validated. Read-only via man -P cat. Options: max_lines (500-10000) with a truncated flag, case-insensitive search with context_lines, and offset. Fatal if man is missing or there is no manual entry. Use when the user asks about flags, syntax, or edge cases.
Inputs: command (required), offset, search, max_lines, +2 more. Returns: errors, command, content, truncated
get_memory_infoA
Returns RAM and swap usage including total, used, and free. Read-only from /proc/meminfo. Fatal only if /proc/meminfo is unreadable. Use for memory pressure checks; get_system_health_check for thresholds.
Returns: free, used, total, errors, +5 more
get_mount_optionsA
Returns mount sources, targets, filesystem types, and options via findmnt. mount_point filters to an exact target. Read-only; fatal if findmnt is missing. Use for mount flags (rw/ro, noexec, etc.); get_disk_info for usage.
Inputs: mount_point. Returns: errors, mounts
get_network_connectionsA
Returns TCP/UDP sockets with state, addresses, process info, and optional reverse-DNS hostnames. Read-only from /proc/net. Params: status and type filters, grouped by PID, max_connections (0-200). resolve_hostnames does network lookups and is slow - keep it off unless needed. For listening sockets pass status=LISTEN; process names are included.
Inputs: type, status, grouped, max_connections, +1 more. Returns: errors, groups, connections
get_network_infoA
Returns per-interface network I/O counters (bytes, packets, errors, drops). Read-only from /proc/net/dev. Fatal only if that file is unreadable. Use for bandwidth and drop analysis; get_network_connections for sockets.
Returns: errors, interfaces
get_power_analyticsA
Returns power state (AC vs battery), discharge rate in watts, battery percentage, and capacity degradation. Read-only from /sys/class/power_supply. Missing sysfs data lands in the errors field. Use for laptop power monitoring.
Returns: errors, ac_online, battery_percent, discharge_rate_watts, +1 more
get_proc_diagnosticsA
Returns deep /proc diagnostics: interrupts, softirqs, vmstat, diskstats, filesystems, version, and slabinfo. The sections param selects a comma-separated subset (empty = all). Read-only; per-section failures land in the errors field. Use for kernel-level debugging.
Inputs: sections. Returns: errors, vmstat, version, slabinfo, +4 more
get_process_fdsA
Lists open file descriptors (files, sockets, pipes) and the total count for a pid. Read-only via /proc//fd. pid is required; an unknown pid returns an error in the errors field, not a failure. Use for fd-leak and resource-hold debugging.
Inputs: pid (required). Returns: pid, name, errors, fd_count, +1 more
get_process_infoA
Returns running processes sortable by CPU, memory, or both with a configurable limit (default 10, max 100). Read-only but heavy: walks /proc and reads smaps for every process, so keep limit low. Per-process read errors are ignored; fails only if the process list cannot be read. Prefer get_top_io_processes for disk I/O ranking.
Inputs: limit, sort_by. Returns: by_cpu, errors, by_memory, processes
get_process_treeA
Returns a flat process list with depth info showing parent-child hierarchy. Read-only from /proc//stat and /proc//comm. When pid is omitted, returns all processes rooted at init (ppid 0). When pid is provided, returns only that subtree. Invalid or missing PIDs return empty nodes with no error. Use to understand process hierarchy and trace spawned children; for per-process resource usage prefer get_process_info.
Inputs: pid. Returns: nodes, total, errors, filtered
get_raid_statusA
Returns software RAID status from /proc/mdstat: devices, levels, sizes, and active/degraded/inactive health. Read-only; an unreadable file is silently swallowed. Use to check array health.
Returns: errors, devices
get_routing_tableA
Returns the kernel routing table: destination, gateway, interface, proto, scope, metric, and MTU. Read-only via ip route show. Use for network path and gateway troubleshooting.
Returns: total, errors, routes
get_security_auditA
Runs a security audit: firewall rules, SSH hardening, SUID binaries, world-writable files, umask, and password policy, with a 0-100 score. Read-only but heavy - scans the filesystem for SUID and world-writable files. Needs root to see everything. Use for hardening reviews, ideally during low load.
Returns: umask, errors, firewall, ssh_hardening, +4 more
get_selinux_apparmor_statusA
Returns SELinux and AppArmor enforcement status. Read-only via getenforce and sysfs/aa-status; never fatal, missing modules report 'not_enabled'. Use for security posture checks.
Returns: errors, selinux, apparmor
get_service_statusA
Returns detailed status of a systemd service (or --user service). name is required and validated. Read-only via systemctl status. Errors from systemctl (e.g. unit not found) appear in the errors field with the raw output; only a missing systemctl is fatal. Use to check why a service failed; get_systemd_units to list states.
Inputs: name (required), user. Returns: pid, name, active, errors, +2 more
get_smart_healthA
Returns SMART disk health: status, temperature, power-on hours, and key attributes via smartctl. device is optional; empty probes all block devices, which is slow with many disks. Requires smartctl and typically root to read /dev. Per-device failures set status to unknown in the errors field; only a missing smartctl is fatal.
Inputs: device. Returns: errors, devices
get_systemd_unitsA
Lists all systemd units and their states. The state param filters by exact match on the Active column ('failed', 'active', 'inactive'). Read-only via systemctl; fatal if systemctl is missing or fails. Use for a full service inventory; get_service_status for one unit's details.
Inputs: state. Returns: units, errors
get_system_health_checkA
Returns an OK/WARNING/CRITICAL verdict derived from memory, disk (partitions at 80%+), load vs core count, and failed systemd units. Returns UNKNOWN when checks are incomplete without an observed warning or critical result. Read-only aggregate; disabled collectors are skipped and omissions or sub-failures land in errors. Use for a first-line status; use get_system_snapshot or targeted tools for detail.
Returns: checks, errors, overall
get_system_infoA
Returns host info: hostname, OS, kernel, architecture, uptime, process count, boot time, virtualization, host UUID, and DMI hardware/BIOS/TPM details. Read-only (gopsutil host.Info plus sysfs). Fails only if host info is unavailable. Use for a single-machine identity summary; for a health verdict use get_system_health_check.
Returns: procs, errors, host_id, os_name, +16 more
get_system_snapshotA
Returns one broad snapshot combining system, CPU, temperature, memory, disk, network, load, top processes, and Docker data in a single call. Read-only. Individual gather failures fall back to zero values in the errors field - never fatal, but the heaviest call in the server. Use as the single entry point for a wide overview; for a pass/fail status use get_system_health_check, and for focused detail prefer the targeted tools.
Returns: cpu, disk, docker, errors, +6 more
get_time_sync_statusA
Returns NTP/Chrony sync state: service, sync status, system and RTC time, stratum, and last offset. Read-only via timedatectl and chronyc; failures are non-fatal (errors field). Use for clock-drift diagnosis.
Returns: errors, stratum, rtc_time, ntp_enabled, +7 more
get_top_io_processesA
Returns processes with the highest disk I/O activity via pidstat -d 1 1 (samples for 1 second). Read-only. If pidstat is missing, returns an error in the errors field with empty results. Use to find which process is hammering the disk; get_disk_io_metrics for per-device totals.
Inputs: limit. Returns: errors, processes
get_user_automationA
Aggregates the current user's scheduled tasks: crontab -l entries and systemd user timers. Read-only; failures such as a missing crontab land in the errors field. Use for user-level automation; get_cron_jobs for system-wide cron.
Returns: errors, cron_jobs, systemd_timers
get_user_infoA
Lists system users from /etc/passwd and /etc/group: username, UID, GID, home, shell, and supplementary groups. search does a case-insensitive substring match. Read-only; a passwd read failure is non-fatal. Use for account inventory and membership checks.
Inputs: search. Returns: users, errors
ping_hostA
Sends ICMP packets and returns latency, packet loss, and response times. host is required and validated; count defaults to 4, timeout to 10s. Read-only but network-bound, blocking up to timeout seconds. Fatal only for an invalid host; a failed ping returns partial results in the errors field. Use for reachability checks; resolve_dns to separate DNS from connectivity.
Inputs: host (required), count, timeout. Returns: host, errors, avg_latency_ms, max_latency_ms, +4 more
resolve_dnsA
Resolves a hostname to IP addresses via the system resolver. hostname is required; a lookup failure returns an error in the errors field. Network-dependent. Use to distinguish DNS failures from connectivity problems; ping_host to test reachability.
Inputs: hostname (required). Returns: errors, hostname, addresses
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 65 tools
Descriptions are unusually disciplined: nearly every tool states its scope and explicitly points to the sibling tool for the adjacent case (e.g. get_disk_info vs get_inode_usage, get_process_info vs get_top_io_processes, get_boot_blame vs get_boot_critical_chain vs get_boot_time). A handful of clusters remain genuinely confusable at a glance — get_system_info / get_system_health_check / get_system_snapshot, and get_docker_info vs get_docker_system_info vs get_docker_disk_usage — but the cross-references resolve them for a careful reader.
All snake_case with a verb_noun pattern, and the get_* prefix is used uniformly for the read/inspect family. The only deviations, ping_host and resolve_dns, still follow verb_noun and are clearly the two active network probes. No camelCase or arbitrary verb mixing.
65 tools is well past the point where an agent can hold the surface in mind, and many are single-file readers (get_file_locks, get_shared_memory_segments, get_logrotate_status, four separate boot-analysis tools) that would be better folded into grouped tools. The breadth of Linux diagnostics is real, but the granularity chosen inflates the count far beyond scope.
For a read-only diagnostics server the lifecycle coverage is very broad: CPU, memory, disk/inode/IO, network, processes/FDs/locks, systemd, boot phases, packages, logs/audit, security posture, hardware/GPU, and a full Docker subtree. Gaps are narrow — no arbitrary log/file content retrieval, no filesystem check or SMART self-test control, and no remediation actions at all (consistent with the read-only design, but a limit if the server is meant for administration).
Maintenance
Related MCP Connectors
Linux server administration over OAuth MCP: shell, Docker, systemd, files and monitoring.
An MCP server giving access to Grafana dashboards, data and more.
MCP server with quote and live cryptocurrency price tools, local and cloud-deployed transports.
Read-only Linux server and Docker monitoring for AI assistants: CPU, memory, disk, network.
Related MCP Servers
- AlicenseAqualityDmaintenanceMCP server for system monitoring, enabling users to ask about CPU, memory, disk, network, and processes on their system.93MIT
- AlicenseNot gradedqualityCmaintenanceA real-time system diagnostics MCP server that gives AI agents live access to CPU, RAM, disk, network, processes, and hardware health metrics, with zero cloud dependency.7Apache 2.0
- AlicenseAqualityDmaintenanceMCP server for remote Linux/Unix server management via SSH, enabling command execution, system monitoring, file operations, and diagnostics through natural language.341MIT
- AlicenseAqualityDmaintenanceEnables AI assistants to perform controlled Linux system administration tasks like reading logs, managing services, cron jobs, WordPress, and executing sandboxed Python code, with strict security constraints.292GPL 2.0