🔴 PSR — Problem Steps Recorder

A Python recreation of the classic Windows Problem Steps Recorder (psr.exe) that Microsoft removed in favor of the Snipping Tool. This version captures a screenshot of the active window on every mouse click, logs what you did, and generates a clean HTML report you can share or archive.

Features

• Automatic screenshot capture on every mouse click just like the original PSR
• Active window capture screenshots are cropped to the foreground window, not the entire desktop
• Click position highlight a red crosshair circle marks exactly where you clicked
• Window title logging records which application was active for each step
• UI element detection identifies the control/element under the cursor (with optional pywinauto)
• Keyboard input tracking logs typing activity with character counts (content is masked for privacy)
• HTML report generation self-contained HTML file with embedded screenshots and a dark theme
• Lightbox viewer click any screenshot in the report to view it full-size; close with ×, click outside, or Escape
• ZIP export optionally bundle the report with separate screenshot PNGs
• Configurable adjust capture delay, output format, and more via CLI flags

Grab your release from the release page

Releases

Requirements ( For development or running as a Python script directly )

Python 3.10+ on Windows.

Core dependencies

pip install pynput mss Pillow pygetwindow

Optional (recommended)

For richer UI element detection (control type, element name, class):

pip install pywinauto comtypes

Usage

# Start recording — press Ctrl+Shift+F9 to stop
python psr.py

# Custom output filename
python psr.py --output my_report

# Adjust minimum delay between captures (default: 0.3s)
python psr.py --delay 0.5

# Don't record keyboard actions
python psr.py --no-keyboard

# Capture full desktop instead of just the active window
python psr.py --fullscreen

# Export as ZIP with separate screenshot files
python psr.py --format zip

# Export both HTML and ZIP
python psr.py --format both

CLI Reference

Flag	Short	Default	Description
--output	-o	steps_report	Output filename (without extension)
--delay	-d	0.3	Minimum seconds between captures (debounce)
--no-keyboard		false	Disable keyboard input logging
--fullscreen		false	Capture entire desktop instead of active window
--format		html	Output format: html, zip, or both
--zip		false	Shorthand for --format both

Stop Hotkey

Press Ctrl+Shift+F9 at any time to stop recording. The report is generated automatically.

You can also press Ctrl+C in the terminal.

Output

HTML Report

A self-contained .html file with all screenshots embedded as base64. Open it in any browser - no server or extra files needed.

Each step shows:

• Step number and timestamp
• Action type (Left Click, Right Click, Keyboard Input)
• Active window title
• Click coordinates
• UI element info (when available)
• Screenshot with click position highlighted

ZIP Archive

When using --format zip or --format both, a .zip is created containing:

• report.html — the report referencing external images
• screenshots/step_001.png, step_002.png, etc.

How It Works

1. Mouse listener (pynput) watches for click events
2. On each click, the recorder:
   • Identifies the foreground window using Win32 API (DwmGetWindowAttribute for accurate bounds)
   • Captures just that window region using mss
   • Draws a red crosshair at the click position
   • Logs the window title and UI element under the cursor
3. Keyboard listener buffers keystrokes and flushes them as a single step after a 1-second pause
4. On stop, all steps are compiled into an HTML report with an embedded lightbox viewer

Comparison with Original PSR

Feature	Original PSR	This Clone
Screenshot on click	✅	✅
Active window capture	✅	✅
Click position highlight	✅ (green border)	✅ (red border)
Window title logging	✅	✅
UI element detection	✅ (MSAA)	✅ (UIA via pywinauto)
Keyboard logging	✅ (masked)	✅ (masked)
Output format	.mht (ZIP)	.html / .zip
Fullscreen option	❌	✅
Comment/annotation	✅	❌ (planned)
Configurable delay	❌	✅

Troubleshooting

"Window: (unknown)" — If pygetwindow can't detect the active window, the recorder falls back gracefully. Installing on Windows with the Win32 API available (default with CPython) gives the best results.

Screenshots are full desktop — The active window detection requires Win32 APIs (ctypes). If you're running in an unusual environment, use --fullscreen and crop manually, or install pywinauto for better window detection.

Double/rapid captures — Increase the debounce delay: --delay 0.5 or higher.

License

GNU GENERAL PUBLIC LICENSE (GPL-3.0)