AI BriefingPickwatermarks-remover: Automatic Removal of AI Watermarks from Claude, Gemini, and OpenAI
guillaumemeyer/watermarks-remover
About the project
A tool that removes hidden provenance marks from text and files generated by AI models. It identifies provenance marks from major vendors such as Claude, Gemini, and OpenAI to manage privacy and content hygiene. It supports a wide range of formats, including text, images, documents, and media files.
It processes marks in three layers. Layer A removes invisible characters such as Unicode, special spaces, and BIDI characters using deterministic scripts. Layer B handles token-sampling-based statistical watermarks through agent rewriting. Layer C deletes C2PA, EXIF, and XMP metadata from various file formats such as PNG, PDF, DOCX, and MP4.
Designed to run without requiring Python installation on the agent host. The skill operates solely as an HTTP client, while actual processing is performed by a service based on the standard library. It integrates as a plugin into various development environments such as Claude Code and Cursor, allowing automatic inspection or removal of marks after file creation.
Useful when ensuring that unwanted metadata is not included before sharing AI-generated content externally or committing it to a repository. It is particularly suitable for consistently managing content provenance marks in environments where multiple AI tools are used interchangeably.
guillaumemeyer/watermarks-remover
A privacy-first app that strips AI watermarks from content you own.
Python
This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.