Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Rewrite AI-generated text to sound natural and human-written. Removes LLM tells — cliché phrases, predictable structure, inflated language, and robotic patterns. Use when editing drafts, emails, articles, or any text that reads like it was written by AI.
.claude/skills/cowork-os-humanizer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-12 | ✓→✗ | ▼ Worse | 9% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -1% | 0% |
| case-13 | ✓→✓ | ＝ Same ✓ | -2% | 0% |
| case-14 | ✓→✓ | ＝ Same ✓ | 17% | 0% |
Rewrite AI-generated text to sound natural and human-written. Removes LLM tells — cliché phrases, predictable structure, inflated language, and robotic patterns. Use when editing drafts, emails, articles, or any text that reads like it was written by AI.
| Name | Type | Required | Description | |---|---|---|---| | text | string | No | The text to humanize (or paste it directly in your message) | | tone | select | No | Target tone for the rewrite |
../humanizer.json.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | fail→fail | 6,840 | 4,022 | -41% | 1 | 1 | 0% | 1,066 | 989 | -7% | 0 | 0 | — |
case-01 | fail→fail | 6,962 | 5,072 | -27% | 1 | 1 | 0% | 1,140 | 1,158 | +2% | 0 | 0 | — |
case-02 | fail→fail | 7,461 | 3,865 | -48% | 1 | 1 | 0% | 1,190 | 953 | -20% | 0 | 0 | — |
case-03 | fail→fail | 9,962 | 5,607 | -44% | 1 | 1 | 0% | 1,504 | 1,207 | -20% | 0 | 0 | — |
case-04 | fail→pass | 5,901 | 4,015 | -32% | 1 | 1 | 0% | 999 | 973 | -3% | 0 | 0 | — |
case-05 | pass→pass | 7,418 | 5,464 | -26% | 1 | 1 | 0% | 1,298 | 1,290 | -1% | 0 | 0 | — |
case-06 | fail→fail | 6,510 | 4,477 | -31% | 1 | 1 | 0% | 1,052 | 1,027 | -2% | 0 | 0 | — |
case-07 | fail→fail | 6,692 | 3,524 | -47% | 1 | 1 | 0% | 994 | 952 | -4% | 0 | 0 | — |
case-08 | fail→fail | 8,079 | 5,762 | -29% | 1 | 1 | 0% | 1,324 | 1,353 | +2% | 0 | 0 | — |
case-09 | fail→fail | 7,670 | 4,461 | -42% | 1 | 1 | 0% | 1,332 | 1,096 | -18% | 0 | 0 | — |
case-10 | fail→fail | 6,536 | 3,338 | -49% | 1 | 1 | 0% | 1,070 | 888 | -17% | 0 | 0 | — |
case-11 | fail→fail | 7,607 | 3,561 | -53% | 1 | 1 | 0% | 1,112 | 968 | -13% | 0 | 0 | — |
case-12 | pass→fail | 5,244 | 3,588 | -32% | 1 | 1 | 0% | 844 | 920 | +9% | 0 | 0 | — |
case-13 | pass→pass | 5,707 | 4,180 | -27% | 1 | 1 | 0% | 983 | 959 | -2% | 0 | 0 | — |
case-14 | pass→pass | 4,519 | 3,218 | -29% | 1 | 1 | 0% | 718 | 841 | +17% | 0 | 0 | — |
case-15 | pass→pass | 6,315 | 5,261 | -17% | 1 | 1 | 0% | 1,010 | 1,241 | +23% | 0 | 0 | — |
case-17 | fail→fail | 6,518 | 4,592 | -30% | 1 | 1 | 0% | 1,029 | 1,041 | +1% | 0 | 0 | — |
case-18 | fail→fail | 8,376 | 4,578 | -45% | 1 | 1 | 0% | 1,222 | 1,067 | -13% | 0 | 0 | — |
case-19 | fail→fail | 7,468 | 3,949 | -47% | 1 | 1 | 0% | 1,240 | 959 | -23% | 0 | 0 | — |
case-20 | pass→pass | 12,807 | 9,417 | -26% | 1 | 1 | 0% | 2,347 | 1,784 | -24% | 0 | 0 | — |
case-21 | pass→pass | 3,394 | 2,130 | -37% | 1 | 1 | 0% | 557 | 711 | +28% | 0 | 0 | — |
case-22 | pass→pass | 2,802 | 3,486 | +24% | 1 | 1 | 0% | 584 | 999 | +71% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of -14 percentage points is the difference between those two pass rates over the 22 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.