Albonire/notebooklm-watermark-remover — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a simple Python tool that removes the NotebookLM watermark from files I already own, mainly PDF slide decks, PowerPoint files, and images like PNG, JPG, JPEG, and WEBP. I do not want it to just slap a white box over the logo. I want it to intelligently rebuild the background so gradients, textures, borders, and transparent areas still look natural.

It should work on a single file or a whole folder, automatically figure out the file type, and save a cleaned copy next to the original with a clear name. For PDFs, add a preview option that only processes the first page so I can test settings before running the full file. A progress bar would be great for bigger jobs.

Please make the command line usage straightforward, include help text, and allow adjusting the watermark area with simple margin settings if detection needs tweaking. If it makes sense, include what is needed to package it as a Windows executable too. Look up current docs online if you need to.

Want more depth? Deep Reverse