Auto-Detection and Auto-Optimization in WinUtil

[gatografiko]

Summary

This proposal explores adding an automatic system profiling and optimization feature to WinUtil.
The goal is to allow the tool to detect the system state (fresh Windows install vs. long-used system) and automatically select appropriate optimizations, while remaining safe, transparent, and fully reversible.

The feature would not replace user control, but rather act as an intelligent preset selector that can optionally apply changes.

---

Motivation

WinUtil already provides:

• Modular tweak and application definitions
• Presets for grouped actions
• A fully automated execution path via config files

What is missing is a decision layer that determines which preset and tweaks are most appropriate for a given system state. This proposal aims to fill that gap.

---

Proposed Feature Overview

Core concept
Add a System Profiler that:

• Analyzes the Windows environment
• Classifies the system into a profile
• Selects the best preset automatically
• Optionally applies the changes

Execution would reuse the existing -Config -Run automation pipeline.

---

Operating Modes

1. AutoDetect + AutoSelect (default)

• Analyze the system
• Automatically select:
  • performance tweaks
  • debloat candidates
  • optimization profile (laptop, desktop, gaming, etc.)
• Present a summary of decisions and reasoning

2. AutoApply (explicit opt-in)

• Only applies changes if explicitly requested
• Prevents accidental destructive actions
• Suitable for scripting and post-install workflows

---

System State Detection (High Level)

Clean / Fresh Windows Install
Detected using:

• OS installation age
• Low number of installed applications
• Minimal non-Microsoft software

Used / Established System
Detected when:

• OS age exceeds threshold
• Significant third-party software is installed
• User workflows are likely present

This distinction allows safe handling of debloat and performance tweaks.

---

Optimization Profiles (Examples)

Based on hardware and environment detection:

• LaptopBalanced
• DesktopPerformance
• GamingLowLatency
• WorkstationStable
• EnterpriseSafe

Each profile maps to a predefined preset.

---

Actions by Scenario

Fresh System

• Apply essential performance tweaks
• Enable recommended settings
• Pre-select OEM / consumer bloatware
• Optional base application install

Used System

• Conservative, non-disruptive tweaks only
• Startup optimization
• Temporary file cleanup
• Safe UI/telemetry adjustments
• Debloat limited to clearly unused consumer apps

---

Implementation Approach

Suggested additions:

• Get-WinUtilSystemProfile.ps1 (private)
• Invoke-WinUtilAutoOptimize.ps1 (public)

Execution flow:

1. Generate a temporary WinUtil config JSON
2. Run using existing automation:

Powershell

-Config generated.json -Run

This minimizes risk and avoids tight coupling with the UI.

---

CLI / Automation Use Case

The feature would support scripted usage, including from bash-compatible shells:

bash

pwsh -NoProfile -ExecutionPolicy Bypass -Command \
  '& ([ScriptBlock]::Create((irm "https://christitus.com/win"))) -AutoDetect -AutoApply'

---

Safety, Reversibility & Governance

• Prefer tweaks with undo support
• Maintain allowlists for enterprise and OEM software
• Provide transparent logging of all actions
• Default to conservative behavior

---

Why This Fits WinUtil

• Aligns with existing modular architecture
• Reuses config-based automation
• Improves usability for new users
• Preserves power-user control
• Scales well for enterprise and scripting scenarios

---

Open Questions for Maintainers

• Should AutoApply be disabled by default?
• What thresholds should define a “clean” system?
• Should enterprise detection override all debloat actions?
• Should presets be community-maintained or core-maintained?

---

Conclusion

This proposal introduces intelligent defaults without removing user agency.
It builds on WinUtil’s strengths and could significantly improve both first-run and long-term optimization workflows.

Feedback and suggestions are welcome.