Skip to content
Get packs

Clean User-Facing Text

Strips hidden Unicode artifacts and tightens prose while keeping facts and voice intact

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

What it does

When installed, the agent can take text you already wrote (articles, reports, docs, emails, product copy) and run a cleanup pass. It removes invisible Unicode characters and lookalike symbols that can slip into pasted text, and it can rewrite prose for flow while being told to keep every fact, number, citation, and required disclosure unchanged. It also includes scripts that score how 'AI-cadenced' a text looks using simple statistics, so the agent can report before and after numbers as a rough gauge.

Great for

  • Writers or marketers polishing their own articles, newsletters, or product copy before publishing
  • Teams that have a documentation or style rule requiring a final text-hygiene pass
  • People who want to find and remove invisible or homoglyph characters from text copied out of chat tools or web pages

Skip it if

  • Hiding authorship or evading AI-content disclosure rules; the skill itself says not to use it for undisclosed authorship evasion
  • Code-only tasks or anything where exact wording must stay verbatim, since rewriting can change meaning if the agent makes mistakes
  • Treating the stylometry score as proof that text is human or will pass any specific detector; the skill says the estimator is not a trained classifier

Try asking

  • “Clean up the invisible characters and smart quotes in this article draft, then tell me the before and after scores”
  • “Polish this product description for clarity without changing any specs or prices”
  • “Audit this report for hidden Unicode characters and list the flagged spots without rewriting anything”
Step-by-step guide: how to use Clean User-Facing TextInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 10 original files, the MIT licence and an attribution note.

    clean-user-facing-text.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip clean-user-facing-text.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip clean-user-facing-text.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

  • Ask for an audit first (detect-only) to see what would be flagged before any rewrite happens
  • Provide your own voice sample only if you own it, and expect the agent to keep your facts and disclosures untouched
  • Read the scores as a gauge, not a verdict; check the output yourself before publishing

What's inside

  • references/detectors.md
  • references/responsible-use.md
  • references/watermark-notes.md
  • references/writing-in-your-voice.md
  • scripts/clean_text.py
  • scripts/common.py
  • scripts/inspect_text.py
  • scripts/score_stylometry.py
  • scripts/text_unicode.py
  • 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.