← All skills · Automation · AI Watermark Checker

AI Watermark Checker — finished output
Runs inClaude Code Codex Cursor Muse Dot Grok Bot Cue OpenClaw Hermes
Needs:The text to check: text.txt (a Source line saying where the text came from, one Text line per paragraph — paste the text as it is, do not tidy it first — and optional Note lines). English text, at least 50 words for a grade; shorter texts are reported but not graded.

Similar Skills

View all skills →

What you get

  • A six-signal table: each signal with its observed value, points, and evidence
  • A grade that is a count of signals — Few, Some, Many, or Strong — never a verdict on authorship
  • The strongest evidence named: the repeated words, phrases, openers, and stock phrases
  • Every gap, input error, and warning named in the report
  • Short texts reported but not graded — under 50 words, no grade is given

Good fit

  • Getting a quick, honest read on whether a draft looks machine-patterned
  • Finding exactly which repetition or uniformity is driving the score
  • Checking a rewrite — re-run and watch the score move

Not for

  • Proving a text is AI-generated — this skill counts statistics, it does not decide authorship
  • Removing watermarks or bypassing detection — detection only
  • Grading short texts — under 50 words, no grade is given

Sample output

An example of what this Skill produces. Sample content, shown for illustration.

Watermark check — sample

Words: 116 | Sentences: 8 | Signals: 10 of 12 | Grade: Strong signals | Gaps: 5 | Errors: 0 | Warnings: 0

Signals: word-repetition 2/2 (top 5 content words 22.4%); sentence-uniformity 2/2 (lengths 13-16 words, CV 0.06); opener-repetition 2/2 ('the' starts 5 of 8 sentences); stock-phrases 2/2 (furthermore, moreover, in conclusion); punctuation-uniformity 2/2 (8 of 8 end with a period); repeated-phrases 0/2.

What this report is not: detection only — a high score does not prove the text is AI-generated, and a low score does not prove a human wrote it.

About this Skill

You say
Check this draft for watermark signals — is it repetitive or uniform enough to look machine-patterned?

How to install

  1. Copy the install prompt into your agent: Read https://skillu.dev/install/agent.md, then install the AI Watermark Checker skill (ai-watermark-checker).
  2. Or download the Skill package from this page after launch and place its folder in your agent’s skills directory.
  3. Call it with a slash command, e.g. /ai-watermark-checker, and describe what you want in plain words.

FAQ

Does a Strong signals grade mean the text is AI-generated?

No. It means many watermark-like signals — repetition and uniformity — were found. A human can write uniform, repetitive text, and a machine can write varied text. The grade counts patterns; authorship is your call, made with evidence beyond this report.

Why is a short text not graded?

Under 50 words the sample is too small for any of the six signals to mean anything — the word-repetition share of a 15-word note is noise. The signals are still counted, but no grade is given.

Get Premium access Browse more skills