Key Results

1. Install Python:

If you don't have Python, download and install it from python.org. Ensure "Add Python to PATH" is checked during installation.

2. Install Required Python Libraries:

Open your terminal (Command Prompt, PowerShell, or Terminal on Mac/Linux) and install the necessary libraries:

For pynput (core keyboard/mouse):

pip install pynput

For PyAutoGUI (image recognition, advanced GUI automation):

pip install pyautogui opencv-python

(opencv-python is needed for the 'confidence' feature in image matching).

If pip isn't found, try: python -m pip install ... or py -m pip install ....

OS-Specific Notes:

• macOS: You may need to grant Accessibility permissions to your terminal/Python/IDE for both `pynput` and `PyAutoGUI` to control the computer. (System Settings > Privacy & Security > Accessibility).
• Linux: For `pynput`, you might need python3-tk and python3-dev. For `PyAutoGUI`'s screenshot features (which it uses internally for image location), you might need scrot (e.g., sudo apt-get install scrot) and potentially python3-pil or python3-pillow.

3. Image Files for Automation (If using image actions):

If your routine includes image-based actions (e.g., "Find Image & Click"):

• Create clear screenshots of the UI elements you want to find (e.g., buttons, icons). Save them as `.png` files.
• Place these image files in a known location. It's often easiest to create an `images` subfolder next to your saved Python script and reference them like `images/my_button.png` in the routine steps. Full absolute paths will also work.
• Ensure image paths in the script match their actual location relative to where the script is run, or use absolute paths.
• Image matching can be sensitive to screen resolution changes, themes, or application UI updates. If scripts fail, re-capture images or adjust confidence.

4. Save and Run the Script:

1. Use the "Download Python Script" button on this page to save the generated .py file.
2. (Optional) Review Script: Open the downloaded .py file in a text editor to inspect the code.
3. Open Terminal: Navigate to the folder where you saved the script (e.g., cd Downloads).
4. Execute Script: Run using python your_script_name.py.
5. Activate Target Window (for hotkey scripts): If not using "Run immediately", click on the application window where you want the actions to occur. The script will then listen for your defined 'Start Key'.

5. Important Notes:

• Stopping Hotkey Scripts: Press your defined 'Stop Key' to terminate the script and its listener. You might need to focus the terminal window running the script for the stop key to register.
• Fail-Safes (PyAutoGUI): PyAutoGUI has a built-in fail-safe: quickly move your mouse to any screen corner to raise an exception and stop a PyAutoGUI-driven script. Consider adding a small `Default Inter-Step Delay` in the "Script Controls" for time to activate this if needed.

6. Using with n8n (Advanced):

The generated Python scripts can be orchestrated by n8n.io using its "Execute Command" node.

• Ensure Python and the required libraries (`pynput`, `pyautogui`, `opencv-python`) are installed in the environment where your n8n instance (or its job executor) runs. Using Python virtual environments is highly recommended.
• In n8n, use the "Execute Command" node. The command would be something like: /path/to/your/venv/bin/python /path/to/your_script_name.py.
• For n8n integration, it's often best to check the "Run routine immediately (no hotkey listener)" option on this page. This makes the script execute its task once and then exit, which is ideal for n8n.
• Passing Data (Conceptual): If you modify the script to accept command-line arguments (e.g., using Python's `sys.argv`), you can pass data from n8n: python script.py "{{ $json.someDataFromN8n }}". The script would then need to parse these arguments.
• **Receiving Data (Conceptual):** If the script is modified to print a JSON string to standard output (e.g., `print(json.dumps({"result": "done"}))`), n8n's "Execute Command" node can capture this in its `stdout` property, which you can then parse with `JSON.parse($json.stdout)`.

This allows you to integrate complex local desktop automations into larger n8n workflows.