Glances System Monitoring Skill

⚠️ MANDATORY SKILL INVOCATION ⚠️

YOU MUST invoke this skill (NOT optional) when the user mentions ANY of these triggers:

"Glances stats", "system health", "CPU usage"
"memory usage", "disk space", "container stats"
"check Glances", "server monitoring", "process list"
Any mention of Glances or system resource monitoring

Failure to invoke this skill when triggers occur violates your operational requirements.

Monitor system health via the Glances REST API (v4). Get real-time stats on CPU, memory, disk, network, sensors, containers, and processes.

Purpose

This skill provides read-only access to system metrics from a Glances server:

CPU usage (total, per-core, load average)
Memory and swap usage
Disk I/O and filesystem space
Network traffic (per interface)
Temperature and fan sensors
Docker/Podman container stats
Process list (top consumers)
System alerts and warnings

All operations are GET-only and safe for monitoring/reporting.

Setup

Start Glances in server mode on the target system:

glances -w --disable-webui  # API only
# or
glances -w                  # API + Web UI

Add credentials to .env file: ~/.homelab-skills/.env

GLANCES_URL="http://localhost:61208"
GLANCES_USERNAME=""  # Optional: leave empty if no auth
GLANCES_PASSWORD=""  # Optional: leave empty if no auth

Variable details:

GLANCES_URL: Glances server URL (default port 61208)
GLANCES_USERNAME: HTTP Basic auth username (optional, leave empty if no auth)
GLANCES_PASSWORD: HTTP Basic auth password (optional, leave empty if no auth)

Commands

All commands output JSON. Use jq for formatting or filtering.

Quick Overview

Get a quick system overview (CPU, memory, swap, load):

./scripts/glances-api.sh quicklook

System Info

./scripts/glances-api.sh system      # Hostname, OS, platform
./scripts/glances-api.sh uptime      # System uptime
./scripts/glances-api.sh core        # CPU core count (physical/logical)

CPU Stats

./scripts/glances-api.sh cpu         # Overall CPU usage
./scripts/glances-api.sh percpu      # Per-core CPU usage
./scripts/glances-api.sh load        # Load average (1/5/15 min)

Memory Stats

./scripts/glances-api.sh mem         # Memory usage (total/used/free/percent)
./scripts/glances-api.sh memswap     # Swap usage

Disk Stats

./scripts/glances-api.sh fs          # Filesystem usage (mount points, space)
./scripts/glances-api.sh diskio      # Disk I/O (read/write bytes/s)
./scripts/glances-api.sh raid        # RAID array status (if available)
./scripts/glances-api.sh smart       # S.M.A.R.T. disk health (if available)

Network Stats

./scripts/glances-api.sh network     # Network interface traffic
./scripts/glances-api.sh ip          # IP addresses
./scripts/glances-api.sh wifi        # WiFi signal (if available)
./scripts/glances-api.sh connections # TCP connection states

Sensors

./scripts/glances-api.sh sensors     # Temperature, fan, battery sensors
./scripts/glances-api.sh gpu         # GPU stats (if available)

Processes

./scripts/glances-api.sh processlist           # Full process list
./scripts/glances-api.sh processlist --top 10  # Top 10 by CPU
./scripts/glances-api.sh processcount          # Process counts by state

Containers

./scripts/glances-api.sh containers            # Docker/Podman containers
./scripts/glances-api.sh containers --running  # Running only

Alerts & Status

./scripts/glances-api.sh alert       # Active alerts/warnings
./scripts/glances-api.sh status      # API status check
./scripts/glances-api.sh plugins     # List available plugins
./scripts/glances-api.sh amps        # Application monitoring (AMPs)

Raw Plugin Access

./scripts/glances-api.sh plugin <name>         # Get any plugin data
./scripts/glances-api.sh plugin <name> <field> # Get specific field

Dashboard (All-in-One)

./scripts/glances-api.sh dashboard   # Comprehensive system overview

Workflow

When the user asks about system health:

"How's the server doing?" → Run ./scripts/glances-api.sh dashboard
"CPU usage?" → Run ./scripts/glances-api.sh cpu
"Memory?" → Run ./scripts/glances-api.sh mem
"Disk space?" → Run ./scripts/glances-api.sh fs
"What's using resources?" → Run ./scripts/glances-api.sh processlist --top 10
"Container stats?" → Run ./scripts/glances-api.sh containers
"Any problems?" → Run ./scripts/glances-api.sh alert
"Temperatures?" → Run ./scripts/glances-api.sh sensors

Output Examples

Quick Look

{
  "cpu": 12.5,
  "cpu_name": "AMD Ryzen 9 5900X",
  "mem": 45.2,
  "swap": 0.0,
  "load": 2.15
}

Memory

{
  "total": 32212254720,
  "available": 17637244928,
  "percent": 45.2,
  "used": 14575009792,
  "free": 1234567890
}

Filesystem

[
  {
    "device_name": "/dev/nvme0n1p2",
    "fs_type": "ext4",
    "mnt_point": "/",
    "size": 500107862016,
    "used": 125026965504,
    "free": 375080896512,
    "percent": 25.0
  }
]

Notes

Glances API v4 is used (default since Glances 4.x)
Requires network access to your Glances server
Default port is 61208
Some plugins may return empty data if not applicable (e.g., gpu without GPU, raid without RAID)
Authentication is optional; configure in Glances with --username and --password
All calls are read-only GET requests

Multiple Servers

To monitor multiple Glances servers, use numbered environment variables in .env:

# In ~/.homelab-skills/.env
GLANCES1_URL="http://server1.local:61208"
GLANCES1_USERNAME=""
GLANCES1_PASSWORD=""

GLANCES2_URL="http://server2.local:61208"
GLANCES2_USERNAME=""
GLANCES2_PASSWORD=""

Then use with server number:

# Server 1 (default)
./scripts/glances-api.sh cpu

# Server 2
SERVER_NUM=2 ./scripts/glances-api.sh cpu

Note: Multi-server support requires script updates (not yet implemented)

Reference

Glances REST API Documentation — Full endpoint reference
Glances Official Docs

🔧 Agent Tool Usage Requirements

CRITICAL: When invoking scripts from this skill via the zsh-tool, ALWAYS use pty: true.

Without PTY mode, command output will not be visible even though commands execute successfully.

Correct invocation pattern:

<invoke name="mcp__plugin_zsh-tool_zsh-tool__zsh">
<parameter name="command">./skills/SKILL_NAME/scripts/SCRIPT.sh [args]</parameter>
<parameter name="pty">true</parameter>
</invoke>