Skip to content
Get packs

Remove AI Marks

Strip invisible AI markers and provenance metadata from text and files via a local cleaning service.

Source Download skill
Repository
guillaumemeyer/watermarks-remover
Stars
25,950
Licence
MIT
Files
7 files

What it does

This skill teaches an AI agent to remove hidden provenance marks that some AI tools leave in their output, such as invisible Unicode characters, C2PA or Content Credentials metadata in images and documents, and certain statistical text watermarks. The agent does not do the cleaning itself: it sends the file to a separate HTTP cleaning service (started by the operator) using curl, then writes the cleaned file back out and summarizes what was removed. The SKILL.md also states that text rewrites are offered as an optional layer and that results should not be presented as proof of human authorship.

Great for

  • Developers who want to remove hidden Unicode characters or leftover metadata from AI-generated text or files they own
  • Creators who want to strip C2PA or EXIF metadata from their own images or documents before sharing
  • Researchers studying how AI provenance marks work and how robust they are

Skip it if

  • Passing off AI-generated work as human-written or evading academic integrity rules, since the skill's ethics guidance warns against this and results should not be marketed as proof of human authorship
  • Removing marks from content you do not own, which may breach terms of service or law
  • Users who cannot run or trust an external cleaning service, since the skill cannot clean anything without it

Try asking

  • “Remove invisible Unicode characters from this notes.md file before I publish it”
  • “Strip C2PA Content Credentials and EXIF metadata from my photo export.jpg”
  • “Clean the AI metadata from this PDF and show me a report of what was removed”
  • “Check whether this text contains watermark detections, then clean it”
Step-by-step guide: how to use Remove AI MarksInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 7 original files, the MIT licence and an attribution note.

    remove-ai-marks.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip remove-ai-marks.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip remove-ai-marks.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Run the /health and /capabilities checks first so you know which backends the service actually has before asking for pixel removal or detection.
  • Inspect a file before cleaning it, and keep a copy of the original since cleaned output is written to a new *.cleaned file by default.
  • Use HTTPS for WATERMARKS_SERVICE_URL when the service is on another host so the API key is not sent in cleartext.

What's inside

  • references/ethics.md
  • references/how-claude-marks.md
  • references/mark-classes.md
  • references/markdiffusion.md
  • references/removal-matrix.md
  • references/vendor-notes.md
  • SKILL.md

Redistributed unmodified from guillaumemeyer/watermarks-remover under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.