Dank System Doctor

AI-powered system health monitor plugin for DankMaterialShell. Tracks CPU, RAM, disk, GPU & temp in real-time, detects pending updates, and uses local AI (Ollama) to diagnose issues and suggest one-click fixes.

Features

• Live metrics — CPU, RAM, disk usage and temperature, updated every 5 seconds; GPU (utilization, VRAM, temp) when nvidia-smi is available, polled every 15s
• Health score — 0–100 with adaptive thresholds (historical baseline + margin) or fixed; status pill shows Healthy/Warning/Critical with reason text
• Compact + detail view — quick-glance gauges and expandable metric cards that fill the window
• Pending updates — detects apt, dnf, pacman, Homebrew; shows count, security/bugfix, ETA; Update now with optional snapshot-first and rollback note
• One-click maintenance — clean caches (apt/dnf/pacman), rotate logs, clear /tmp, TRIM SSD, repair broken packages; safe-mode prechecks (free space, battery/AC, snapshot)
• Snapshot/restore guardrail — Timeshift, Btrfs, or ZFS snapshot before destructive actions
• Process monitor — top CPU consumers with mini progress bars
• Log viewer — recent journalctl error entries, polled every 30 seconds
• AI Doctor — summarized context and triage playbooks (high CPU by process, memory leak, disk I/O); one-click "Apply Fix" with root via pkexec

Screenshots

Overview	Processes

Logs	Update & Care

Overview — metric cards with live progress bars and quick-glance gauges · Processes — top CPU consumers · Logs — recent journal errors · Update & Care — updates status and maintenance actions

Requirements

Dependency	Purpose
DankMaterialShell ≥ 1.4.0	Shell framework
Ollama	Local AI backend
bash, ps, free, df, journalctl	System metric collection (standard on any Linux system)

Installation

Via DMS Plugin Manager (recommended)

1. Open DankMaterialShell Settings → Plugins → Install
2. Enter the repo URL: NordicsSys/DankSystemDoctor
3. Enable the plugin and add it to your bar

Manual

cd ~/.config/DankMaterialShell/plugins
git clone https://github.com/NordicsSys/DankSystemDoctor dankSystemDoctor

Then add to plugin_settings.json:

"dankSystemDoctor": {
    "enabled": true,
    "ollamaModel": "llama3.2",
    "cpuThreshold": 85,
    "ramThreshold": 85,
    "diskThreshold": 90,
    "logInterval": 30
}

And add to your bar config in settings.json:

{ "id": "dankSystemDoctor", "enabled": true }

Ollama Setup

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull a model (choose one)
ollama pull llama3.2        # recommended — best balance
ollama pull llama3.2:1b     # fastest, smaller
ollama pull mistral         # good for technical analysis
ollama pull deepseek-r1:7b  # strong reasoning

Ollama runs as a systemd service automatically after install. The plugin connects to http://localhost:11434.

Settings

Setting	Default	Description
Ollama Model	llama3.2	Model name for AI diagnostics
CPU Threshold	85%	Health score penalty above this
RAM Threshold	85%	Health score penalty above this
Disk Threshold	90%	Health score penalty above this
Log Interval	30s	How often to poll journalctl

How the Health Score Works

Starts at 100 and loses points when:

• CPU > threshold → −20 (threshold can be adaptive: baseline average + margin)
• RAM > threshold → −20
• Disk > threshold → −15
• Temperature > 80°C → −15 (> 90°C → −30)
• Journal errors found → −10 (many errors → −15)
• GPU temp > 90°C (if available) → −15

Healthy (80–100) · Warning (50–79) · Critical (0–49). The status pill shows a short reason when not healthy.

License

MIT