Skip to content

Latest commit

 

History

1,207 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PC Workman HCK

Your PC finally has someone who speaks its language.

Get it from the Microsoft Store Version Status Python License Platform

After 12 months of work: on the Microsoft Store. Just click Install.

Get it from Microsoft

Or download the latest release, one ZIP, no installer, no account.

pcworkman.dev - website, guides, and the build-in-public blog (Monday / Wednesday / Friday).


Most monitoring tools give you numbers. PC Workman gives you answers.

Ask "why is my PC slow right now?", and get a real explanation, not just a percentage.
Ask "is cs2.exe a virus?", get an instant process identity check.
Ask "which game pushes my hardware the hardest?", get a thermal signature breakdown.

127 AI intents. The assistant runs 100% on your machine, no API key, no cloud LLM. Just your PC talking to you.

It learns your machine: 82°C is normal while you game but critical at idle, judged against your own history, not a generic 85°C line. Drop an in-game overlay that shows only what you choose, hit ⤢ for a full-screen control center, and watch the Learning Center fill up as it gets to know your hardware. Built by one person, in public, on real hardware.


What makes it different

Traditional tools PC Workman HCK
CPU: 87% "CPU at 87% - Chrome and Electron processes, consider closing Discord"
Static charts Pan/zoom charts: click any point, see pinned tooltip with baseline context
Dumb thresholds Learns your hardware's normal temperatures per workload - gaming vs idle
Voltage: raw numbers SPC control limits - flags deviations, not just ATX spec crossings
No context Remembers patterns, compares today vs your 7-day average
Manual checks Proactive alerts: voltage anomaly? temperature spike? process appeared? Auto-pushed
English only Polish + English, auto-detected per message

Quick Start

Windows Users (Easiest)

1. Download PC_Workman_HCK_1.8.5.zip from Releases
2. Extract the folder anywhere
3. Run "PC Workman HCK.exe" and you are done.
   (keep the _internal folder next to it - that's the runtime)

Get Latest Release

Developers

git clone https://github.com/HuckleR2003/PC_Workman_HCK.git
cd PC_Workman_HCK
pip install -r requirements.txt
python startup.py

Full setup guide: GETTING_STARTED.md

Features

Core Monitoring

  • Real-time CPU, GPU, RAM tracking
  • Network bandwidth per-application
  • Process identification and labeling
  • Temperature monitoring with trends
  • Historical data logging (daily, weekly, monthly)

Maximized View Mode (redesigned in 1.7.9)

  • One click (⤢) turns the compact 1160×575 dashboard into a full-screen control center
  • Symmetric layout: TOP 8 user processes left, TOP 8 system processes right, chart hub in the middle
  • Hardware cards grow into mini-charts with the component name drawn inside the chart corner
  • hck_GPT chat scales with the window: +12% default height, +35% in chat-Maximize
  • Gaming-HUD chart tooltip: hover to inspect any bar at 72% opacity, click to pin, the pin follows the live buffer and its age keeps ticking

Upgrade Readiness (new in 1.8.5)

  • Type a planned purchase ("i5 11400F", "RTX 4070", "DDR5 6000") and get a clear verdict: fits, fits after a BIOS update, or wrong socket with what a swap costs you (board, RAM, cooler mount)
  • 320-entry offline hardware library: 174 desktop CPUs, 79 GPUs, 58 chipsets with per-generation support - it knows the traps (B460 cannot run 11th gen, LGA1151 v1 vs v2, B550 blocks Ryzen 1000/2000)
  • Quick-pick suggestions filtered to what YOUR chipset actually runs; entry buttons at each part in My PC > Components
  • Ask it in chat too: "will a 5800X3D work on my board?", "what RAM fits?" - and the Optimization Center's Upgrade Advisor reads your own 14-day load history to say what is worth buying first

Fan Dashboard (rebuilt in 1.8.5)

  • Curve chart with a heat gradient that brightens to red as you raise values, dual % / RPM axes, monotonic drag-safe points and a live temperature marker
  • The APPLY rule: chart and sliders are a draft (cards preview in amber); Apply persists the plan and re-locks the chart behind a hover padlock
  • hck_GPT [AI] button on the chart: a temperature health check from your learned history, then the learned "hck_GPT - AI" profile chat can apply for you

Intelligence (hck_GPT)

  • 127 intents across 8 categories (hardware, diagnostics, performance, why, optimization, security, fun, small talk) + gaming/battery/upgrade/privacy
  • Hybrid Engine: rule-based responses for known intents, Ollama LLM for open-ended questions, 100% offline, no API key needed
  • Bilingual: Polish and English, auto-detected per message
  • Session memory: conversation context, CPU/RAM trend buffers, cross-response data store
  • Proactive monitor with DeepMonitor integration: background daemon watches CPU/GPU temps, RAM, throttle, disk on all drives; pushes alerts automatically; banner shows live temps
  • Conversation flow: greeting, thanks, "more info", "what should I do" all handled naturally with context-aware routing
  • Local insights engine, habit tracking, anomaly awareness, teasers from Stats Engine
  • SQLite user knowledge base (hardware profile, usage patterns) at AppData/Local/
  • Background hardware scanner (psutil + WMI, CPU model, GPU, VRAM, mobo, RAM speed, disk model)
  • Chat panel nav links: clickable [→ Page] tokens route directly to app pages
  • _followup() pool system: 8 keys, every response ends with a contextual next-question hint

Monitoring & Alerts (significantly extended in 1.7.8)

  • Interactive charts (ui/components/interactive_chart.py): pan (drag), zoom (scroll wheel around cursor), reset (double-click), crosshair + live value bubble, click-pin persistent tooltip with anomaly reason and baseline deviation
  • Minimap strip below each chart, full data range with drag-to-navigate selection window
  • Thermal Baseline Engine (core/thermal_baseline.py): learns CPU temperature norms per workload context (idle / light / medium / heavy / gaming) with a true Welford online accumulator, running per-bucket stats that accumulate over the whole install life and survive 90-day snapshot pruning, not a fixed window. Chart baseline band shows the learned range, not a window average.
  • Voltage Rail Analyzer (core/voltage_analyzer.py): SPC on 12V / 5V / 3.3V rails using Median + MAD. Nelson Rules 1/2/3/5 (isolated spike, cluster, sustained deviation, trend). 12V GPU-transient suppression. Anomaly decay: pattern repeats ≥5× → "your normal".
  • Learning Center: per-workload thermal training progress + learned ranges, per-rail voltage SPC baselines, overall %, live PSU health score, and a ↻ Rebuild self-check.
  • hck_GPT integration: _check_voltage_rails() fires voltage_spike / voltage_trend proactive alerts (bilingual, budget-controlled). format_for_chat(lang) on VoltageAnalyzer.
  • Dashboard chart tooltip: hover any bar → translucent detail box (CPU/RAM/GPU% + sample age) next to the cursor. Click to pin, the tooltip docks to its bar, the age ticks live, and the PIN strip mirrors the hck_GPT TIP/HOT style. Click anywhere to unpin.

DeepMonitor (new in 1.7.6)

  • ttk.Treeview sensor table with 4 aligned columns (Sensor / Value / Min / Max)
  • Type-specific row background tints: temperature = blue-night, utilization = indigo
  • Action bar: Save Data (.txt/.csv), Pause, Reset min/max
  • Sub-section headers color-coded by metric type

MAP OF COMPONENTS (new in 1.7.6)

  • 2.5D isometric view of your PC rendered via Pillow (2× SSAA, LANCZOS downscale)
  • Desktop PC mode: case, mobo, CPU + heatsink, GPU, RAM, SSD, PSU, fans, cables
  • Laptop mode: open chassis with mobo, fans, GPU, battery, screen, keyboard
  • Components color-shift green → amber → red based on live heat/load; hot components pulse
  • Hover over any component for a tooltip with live stats
  • Auto-refresh every 3 seconds via background thread

Live Guide (new in 1.7.2)

  • Interactive 3-step spotlight overlay (ui/guide/live_guide.py) launched from Guide page
  • Windows -transparentcolor technique: dim full screen, punch transparent hole over target widget
  • Step 1: main realtime chart + time-filter buttons; Step 2: left/right nav buttons with per-button descriptions; Step 3: hardware cards + session averages
  • Floating info card with accent bar, step dots, DALEJ / Zakończ button; ESC or ✕ to dismiss

First Setup & Drivers (new in 1.7.2)

  • Health score gauge (0–100 arc), computed from driver ages and startup count
  • 4 driver health cards: GPU, Audio, Network, USB, real data from Windows registry
  • Driver freshness bar per card; status badges: CURRENT / 6+ MONTHS / Xmo OLD
  • Startup program list (registry HKCU/HKLM Run keys), 6-item setup checklist with persistent state
  • Quick Actions: Windows Update, Device Manager, Services, Task Scheduler, System Info, MSConfig

Startup Manager

  • Reads every startup source: HKCU/HKLM/HKLM32 Run keys, Startup folders, Task Scheduler logon/boot tasks, and Microsoft Store (UWP) startup apps, so GPU Tweak, ShareX, LinkedIn, MSI Center and the like finally show up
  • Knowledge base of common programs with impact rating (High/Medium/Low) and recommendation
  • Reversible enable/disable per source (registry removal · schtasks · UWP state); locale-independent task scan works on non-English Windows; choices persist to data/cache/startup_prefs.json

Services Manager - mode configurator

  • Catalogue of 40+ Windows services in 4 categories (Essential locked / Recommended / Optional / Likely Unnecessary), plus enumeration of every installed service
  • Guided Quick setup strip: plain questions ("Do you use Bluetooth?") build your custom profile
  • Per-service G / E / M chips assign each service to the Gaming · Economy · MANAGER modes, one source of truth, synced live with the Features mode buttons (settings/turbo_services.json)
  • Stop / Start / Restart per service; admin detection; all changes logged to data/logs/service_changes.log

Interface

  • Modern dashboard (Apple-inspired design)
  • Ultra-compact information density
  • Color-coded process lists
  • Interactive charts and metrics
  • Click-to-investigate functionality
  • Process tooltips on TOP 5 panels, hover any process name for instant library lookup

On the roadmap

  • Monitoring & Alerts layout polish pass
  • Per-rail voltage history export (CSV)

Architecture

Modular, scalable design:

PC_Workman/
├── core/
│   ├── monitor.py             # psutil snapshot every 1s (background thread)
│   ├── scheduler.py           # drives aggregation ticks
│   ├── thermal_baseline.py    # workload-aware temp learning - Welford accumulator, 5 buckets
│   ├── voltage_analyzer.py    # SPC on 12V/5V/3.3V - Median+MAD, Nelson rules 1/2/3/5
│   └── hibernation_manager.py # SetPriorityClass + NtSuspendProcess for Turbo Mode
├── hck_gpt/
│   ├── engine/        # Hybrid Engine: rule routing + Ollama LLM client
│   ├── intents/       # Intent parser, ML classifier (Naive Bayes), vocabulary, lang detect
│   ├── memory/        # Session memory, user knowledge (SQLite), proactive monitor
│   ├── context/       # System context builder, hardware scanner
│   ├── responses/     # Bilingual response builder (facade + 9 mixins, 109 handlers)
│   ├── chat_handler.py
│   ├── insights.py    # InsightsEngine (habits, anomalies, teasers)
│   └── panel.py       # Chat panel UI
├── hck_stats_engine/  # SQLite pipeline: minute/hourly/daily/monthly stats
├── ui/
│   ├── windows/       # Main window modes (expanded, minimal)
│   ├── guide/         # Interactive spotlight guide (LiveGuide, 3-step tour)
│   ├── components/    # interactive_chart.py, pc_map.py (2.5D isometric), LED bars, tooltips
│   └── pages/         # Full-page views (monitoring, fan control, startup, services)
├── data/
│   ├── logs/          # CSV logs (raw, hourly, daily, weekly, monthly)
│   ├── cache/         # Runtime cache & process patterns
│   └── hck_stats.db   # SQLite long-term storage (WAL mode)
└── tests/             # Unit tests

Design principles:

  • Dynamic component registry (auto-registration)
  • Direct inter-module communication
  • Designed for future expansion
  • Educational value (demonstrates Python best practices)

What's New [1.8.7] - 2026-08-08 - CURRENT

hck_GPT - guided flows and a conversation that holds together

  • Multi-step guides are now an engine rather than one hand-written handler, so every flow understands "next", "yes", "skip" and "stop" for free. Four ship: optimize, cooling, desktop repair and upgrade planning. The optimize flow measures first, walks startup and services, asks before touching RAM, then verifies against the numbers it took at the start.
  • A conversation director (11 handlers) keeps a diagnosis coherent when a real person wanders: correcting the subject ("I meant the GPU"), supplying a detail that was missing a moment ago, coming back after trying the advice, or asking why the assistant is confident.
  • A semantic routing layer covers the relationships a keyword parser drops: process plus close plus safe is a different question from process plus memory. Out-of-domain questions are refused openly instead of being forced into the nearest intent.

The 23 published guides are reachable from inside the app

  • Hard questions now end with an offer to the guide that answers them in full: 22 guides mapped to 54 intents, offered once per guide per session, after the live answer and never instead of it. Guide links are amber because they open a browser; in-app navigation stays purple.
  • The Guide page carries a banner to the same library. Every target is checked by a test against the real files in both languages.

Optimization Receipts cover TURBO

  • TURBO suspended processes and stopped services without ever saying what it gained. Both now print a before and after receipt with the count of what was touched, measured after the action has had time to work rather than 20 seconds in.

Safety and accuracy

  • The mini-antivirus stopped flagging the Print Spooler. A process-library note meaning "heavy or bloatware-class" was escalating the security verdict, so spoolsv.exe and dllhost.exe read as "caution" despite a valid Microsoft signature. Masquerade, homoglyph and miner detection are unchanged and now pinned by tests.
  • The in-game overlay checks the PID before the name, so the app can no longer be asked to freeze itself. Ultrawide screens no longer get a window taller than the screen. The diagnostic console explains itself when Windows Terminal will not let it hide.
  • Process library grew from 485 to 521 entries, with vendors read from the Authenticode signature rather than assumed. Test suite from 229 to 316.

What's New [1.8.5] - 2026-07-21 (previous)

Upgrade Readiness - part compatibility before you buy

  • Type a planned purchase (a CPU, GPU or RAM kit) and get a clear verdict: fits, fits after a BIOS update, or wrong socket with what the swap actually costs you (board, RAM, cooler mount). Fully offline, 188 CPUs and 84 GPUs, 58 chipsets with per-generation support - it knows the traps a spec sheet hides, like a B460 board refusing an 11th-gen chip that shares its exact socket.
  • Live autocomplete: start typing a model and pick it from a list. Entry buttons sit at each part in My PC > Components, and hck_GPT answers the same questions in chat ("will a 5800X3D work on my board?").

Fan Dashboard - rebuilt

  • Heat-gradient curve that brightens to red as you raise it, dual % and RPM axes, drag-safe points that cannot cross, live fan rings that follow real sensors, and an APPLY rule so the chart draft and the applied plan are two clear states. The hck_GPT [AI] button runs a temperature health check from your learned history and can apply a learned fan profile.

Monitoring & Learning

  • Voltage learning now covers CPU VCore and GPU core, not just the board rails. Learned-baseline anomalies land in the Events log with context ("82°C in gaming workload, normal 57-71°C"). The Learning Center type and layout were cleaned up and fully localized.

Performance - the app is faster and leaner

  • My PC used to take about 2 seconds to open. It is now built once and kept alive, so re-entry is 1-17 ms. The real cause was two blocking wmic calls on the UI thread; both now read the hardware identity warmed at startup. The sidebar builds its subitems lazily, cutting steady-state window init to roughly 190 ms.
  • One version source (utils/app_version.py) drives every title, badge and the build name, guarded by a test. A freeze watchdog and a global error log now leave evidence when something stalls. Per-process CPU is on the whole-machine scale (one busy thread no longer reads "100%" on a 12-thread PC). The process library grew from 373 to 485 definitions.

Under the hood

  • The 6,533-line hck_GPT response builder is now a facade over eight mixins, guided flows and a response ledger were added (96 intents), and the sidebar was restructured. Test suite grew from 21 in June to 194.

What's New [1.8.0] - 2026-06-22

Patched - 2026-06-28

  • Services Manager rebuilt as a real configurator: one Wyłącz/Włącz per service feeds a single operator drawer at the bottom of the page, and Zatwierdź applies the whole batch at once, no more per-row dialogs or truncated service names. SZCZEGÓŁY expands the queued list inline.
  • Hardware detection fixed for Windows 11 24H2+ (build ≥ 26100), where wmic.exe was removed, My PC → Components now fills in CPU / GPU / RAM / motherboard / disks via a PowerShell CIM fallback.
  • Fixed a Features-page crash, made expanded feature cards widen for readability, scoped Services Manager scrolling to the page, and hardened sc / PowerShell output decoding so unusual service names can't crash a reader thread.

Smart Learning - engines wired in, and they accumulate

  • hck_GPT now answers temperature with the learned, workload-aware verdict instead of a fixed 85°C cutoff: 82°C reads normal under a gaming load but critical at idle. "voltage check" got its own real handler (was silently aliased to the temperature one). The chat handler imported neither learning engine before, months of learning it couldn't reach.
  • The proactive monitor judges CPU temperature against the learned per-workload baseline (z-score), falling back to fixed thresholds until a bucket is trained, so it stops crying wolf during normal gaming. Elevated-but-safe goes out as a 💡 TIP, not an alarm.
  • Thermal baseline is a real Welford accumulator now: each pass folds only the newest snapshots into a running per-bucket {n, mean, M2}, so learning accumulates for the life of the install and survives 90-day pruning, and a continuous tick in the proactive loop keeps it learning while the app runs.
  • Learning Center in Monitoring & Alerts shows live what was learned: per-workload thermal progress + ranges, per-rail voltage SPC baselines, overall %, PSU health score, and a ↻ Rebuild self-check.
  • Voltage rail health now counts genuine Nelson-rule anomalies (after GPU-transient suppression + recurrence decay), not the ~1.2% Gaussian tail, a healthy rail no longer reads "critical" once enough samples pile up.
  • hck_GPT volunteers two positive learning notes: a one-time 💡 when a workload reaches full calibration ("I now judge temperature against YOUR normal"), and a 💡 "new normal" when a recurring voltage blip becomes your baseline. Both deduped so they never nag.

GAMING - In-Game Overlay

  • New GAMING / In-Game tile in My PC: a translucent always-on-top HUD that floats over borderless / windowed games without stealing focus. Left/right-click moves it between the four corners.
  • Real HUD table, one row per component (CPU / GPU / RAM / 12V), FPS as a side box, with live values.
  • A form-style configurator: 3 presets, or Create Custom where each field is a ▼ dropdown to pick the metric per row, plus a style panel (size / theme / opacity). Live preview matches the overlay 1:1.
  • Live FPS read from RTSS (RivaTuner / MSI Afterburner), no admin, no DLL injection; shows "-" when RTSS isn't running. Per-pixel transparency is on the way.
  • Game launch greetings: a one-second corner toast when a known game starts, now bilingual (PL/EN) with random variants across 40+ games (Planet Zoo, Terraria, Minecraft, Helldivers 2, GTA V, Hades…).

Startup Manager - sees everything

  • Now enumerates Task Scheduler logon/boot tasks and Microsoft Store (UWP) startup apps, not just Run keys + Startup folders. GPU Tweak, ShareX, LinkedIn, MSI Center and others finally appear, each with a reversible enable/disable and a source badge (⏰ Task · ⊞ Store).

Services Manager - configurator + MANAGER mode

  • Rebuilt as a configurator: per-service G/E/M chips assign services to Gaming, Economy or the new custom MANAGER mode, plus a guided question strip. All modes share one config, synced live with the Features buttons.
  • New MANAGER mode in Features (white chip) with a click-through that jumps to the Services Manager.

hck_GPT

  • Process Suspect Guard mini-AV: author (Authenticode) verification, typosquat/homoglyph detection (svhost, ciaude…) and masquerade checks, wired into "virus check" and process identity.
  • Natural-language routing overhaul (everyday phrasings hit the right intent) and purple highlighting of hardware names in chat.
  • Four new data-driven answers: "what should I upgrade?" (the real bottleneck from your own load + temperature history), "do you spy / what do you collect?" (honest, local-only, links to Stability Tests), greetings that name your favourite app ("Fancy CS2 again today?"), and "what starts with Windows?" (your real startup list, links straight to the Manager). Vocabulary now 92 intents.

Earlier releases

Every release from v1.5.7 onward is documented in CHANGELOG.md, including v1.8.5 (Upgrade Readiness, Fan Dashboard rebuild), v1.8.0 (Smart Learning, Microsoft Store) and the full v1.7.x line.

Project Structure

HCK_Labs/PC_Workman_HCK/
├── core/
│   ├── monitor.py           # Background-threaded system monitoring
│   ├── logger.py            # File logging system
│   ├── analyzer.py          # Data analysis & trends
│   ├── scheduler.py         # Background scheduler
│   ├── process_classifier.py # Process categorization (Gaming/Browser/Dev/etc.)
│   └── process_data_manager.py # Process tracking & statistics
├── hck_gpt/
│   ├── chat_handler.py      # Command routing (stats, alerts, insights, etc.)
│   ├── insights.py          # Local InsightsEngine (habits, anomalies, teasers)
│   ├── panel.py             # Chat panel UI (gradient banner, ticker, greeting)
│   ├── report_window.py     # Today Report Toplevel (chart, processes, alerts)
│   └── services_manager.py  # Windows services optimization
├── hck_stats_engine/
│   ├── db_manager.py        # WAL-mode SQLite, thread-local connections
│   ├── aggregator.py        # Minute/hourly/daily/monthly aggregation
│   ├── process_aggregator.py # Per-process CPU/RAM tracking
│   ├── query_api.py         # Range queries with auto-granularity
│   ├── events.py            # Spike/anomaly detection
│   └── constants.py         # Retention config (7d/90d/forever)
├── ui/
│   ├── windows/
│   │   ├── main_window_expanded.py  # Full dashboard (980x575)
│   │   └── main_window.py           # Minimal mode
│   ├── components/
│   │   ├── charts.py, led_bars.py, yourpc_page.py, ...
│   └── pages/
│       ├── monitoring_alerts.py     # Time-Travel Statistics Center
│       ├── fan_control/             # Fan curves & hardware
│       ├── startup_manager.py       # Startup programs manager (new)
│       ├── services_manager.py      # Windows services + TURBO (new)
│       ├── optimization_services.py # Optimization Center
│       └── first_setup_drivers.py  # Driver health & checklist
├── data/
│   ├── logs/                # CSV logs (raw, hourly, daily)
│   ├── cache/               # Runtime cache
│   └── hck_stats.db         # SQLite long-term storage
├── tests/
├── CHANGELOG.md
├── requirements.txt
├── startup.py
└── import_core.py

Installation

Requirements

  • Python 3.9+ (or use .exe)
  • Windows 10+ (Linux/Mac support coming)
  • RAM: 200MB minimum
  • Disk: 300MB (if using .exe installer)

From Source

# Clone repository
git clone https://github.com/HuckleR2003/PC_Workman_HCK.git
cd PC_Workman_HCK

# Create virtual environment (recommended)
python -m venv venv
.\venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run
python startup.py

From .exe

Download from Releases → Double-click → Done

Usage

First Launch

  1. Dashboard opens showing real-time metrics
  2. Give it 5 seconds to collect initial data
  3. CPU/RAM/GPU bars populate
  4. Click tabs to explore features

Main Tabs

  • Dashboard - Real-time overview
  • Your PC - Hardware health & component status
  • Fan Control - Custom fan curves (advanced)
  • Network - Per-app bandwidth usage
  • Gaming - Game-specific analytics

Understanding the Data

  • Green (0-30%) - Normal operation
  • Yellow (30-60%) - Moderate load
  • Orange (60-85%) - Heavy load
  • Red (85%+) - Critical

Click any process to see more details.

Data & Privacy

What's Collected

  • CPU/GPU/RAM usage (on your device only)
  • Process names (to identify running applications)
  • Temperature readings (from hardware sensors)
  • Network usage (local tracking)

Where It's Stored

  • Local only: /data/logs/hck_stats.db (SQLite) + learned baselines in /data/cache/
  • Never personal: no files, keystrokes, browsing, or content, ever
  • You control the network: every outbound connection goes through one gate in Settings; turn it off and the app makes zero connections
  • You control the data: delete /data/ anytime to start fresh

Privacy Assurance

  • All monitoring runs locally on your machine
  • Open source, the code is auditable
  • Network access is optional and off-able in Settings (off = firewall-verifiable zero traffic)

Versioning

Version Status Key Features
v1.0.0 Released Basic architecture
v1.0.6 Stable First working UI
v1.3.3 Released hck_GPT integration
v1.4.0 Released System tray, enhanced UI
v1.5.7 Released Modern dashboard, hardware monitoring
v1.6.3 Released Fan dashboard, menu system, .exe
v1.7.1 Released Stats Engine v2, Time-Travel, Monitoring
v1.7.2 Released Startup/Services Manager, Optimization Hub, hck_GPT AI layer, Hybrid Engine (Ollama), bilingual, EXE build
v1.7.3 Released Live Guide, hck_GPT AI quality (followups, help rewrite, optimization live), session data store, WMI scan, nav links
v1.7.4 Released Optimization Center redesign (2-col grid, expandable cards, Turbo PP creation, Weekly Report, LIVE NOW), dashboard button restyle
v1.7.5 Released hck_GPT 13 new intents (community requests), 4 MEGA features (Time-Windowing, No-AI-Slop, Time-Travel Debug, Micro-Bench), process library 104->241
v1.7.6 Released DeepMonitor rewrite (Treeview), MAP OF COMPONENTS (2.5D isometric), hck_GPT Wave 2 (6 intents, 82 total), font system 100% coverage
v1.7.7 Released Ghost Driver Detection (pnputil), RAM Flush exclusion menu, SEE EVERYTHING/OUTDATED driver views, HOT strip stability, MAP fix
v1.7.7-patched Released UI/UX fixes: Startup Manager redesign, Services Manager stop/start logic, Drivers page readability, compact headers
v1.7.8-monitoring Released Thermal Baseline Engine, Voltage Rail Analyzer (SPC + Nelson rules), interactive pan/zoom charts, proactive voltage alerts, 6-module code-quality pass
v1.7.9 Released Maximized View Mode redesign, gaming-HUD chart tooltip, hck_GPT on MY PC / Fan tabs, 1326 dead lines removed, 10+ resource leaks fixed
v1.8.0 Released Smart Learning (Welford accumulator, workload-aware temps, voltage SPC), GAMING in-game overlay + configurator, live FPS via RTSS, 40+ game greetings, 4 new hck_GPT intents, Process Suspect Guard
v1.8.1 Released Data machine, anti-cheat guard, learning v3, always-on AUTO
v1.8.2 Released Critical freeze fix, admin elevation, scaling, hck_GPT expansion
v1.8.5 Current Upgrade Readiness (offline part compatibility), Fan Dashboard rebuild, My PC 2s -> 17ms keep-alive, voltage learning (VCore/GPU), one version source, builder split, 21 -> 194 tests
v2.0.0 Q4 2026 Long-term drift, Smart User Activity, Tools & Utils

Contributing

For Users

For Developers

  • We welcome pull requests
  • Follow existing code style
  • Include tests for new features
  • Update documentation

System Requirements

Minimum:

  • Python 3.9+
  • Windows 10
  • 200MB RAM
  • 300MB disk space

Recommended:

  • Python 3.11+
  • Windows 11
  • 500MB+ RAM
  • SSD storage

For Gaming Analytics:

  • NVIDIA/AMD GPU drivers updated
  • DirectX 12 compatible system

Documentation

About

Marcin Firmuga | Software Engineer

Building PC Workman in public, physical work by day, code by night.

Part of HCK_Labs initiative.

License

MIT License © 2025 HCK_Labs / Marcin Firmuga Free for personal and commercial use. Attribution appreciated.

Ship what you have. Improve it later. 💙

About

System monitor with offline AI (82 intents, 9-layer routing) that learns YOUR PC. TURBO optimization, thermal baselines, voltage SPC, ghost driver detection, 373 process definitions. optimization features, 11 months solo. No cloud, no API, your data stays local. MIT licensed. - Marcin HCK Firmuga - Build in Public (3 days series every week)

Topics

Resources

Contributing

Security policy

Stars

77 stars

Watchers

25 watching

Forks

Releases

Sponsor this project

Packages

Used by

Contributors

Languages