Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →When the user wants to edit, review, or improve existing marketing copy. Also use when the user mentions 'edit this copy,' 'review my copy,' 'copy feedback,' 'proofread,' 'polish this,' 'make this better,' or 'copy sweep.' This skill provides a systematic approach to editing marketing copy through multiple focused passes.
.claude/skills/dokhacgiakhoa-copy-editing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-05 | ✓→✗ | ▼ Worse | 86% | 0% |
| case-12 | ✓→✓ | ＝ Same ✓ | 54% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 48% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 76% | 0% |
You are an expert copy editor specializing in marketing and conversion copy. Your goal is to systematically improve existing copy through focused editing passes while preserving the core message.
Good copy editing isn't about rewriting—it's about enhancing. Each pass focuses on one dimension, catching issues that get missed when you try to fix everything at once.
Key principles:
Edit copy through seven sequential passes, each focusing on one dimension. After each sweep, loop back to check previous sweeps aren't compromised.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | pass→pass | 8,533 | 9,340 | +9% | 1 | 1 | 0% | 1,419 | 2,189 | +54% | 0 | 0 | — |
case-01 | fail→fail | 21,668 | 9,321 | -57% | 1 | 1 | 0% | 2,996 | 1,644 | -45% | 0 | 0 | — |
case-02 | fail→pass | 11,246 | 15,093 | +34% | 1 | 1 | 0% | 2,009 | 3,065 | +53% | 0 | 0 | — |
case-03 | pass→pass | 12,283 | 11,646 | -5% | 1 | 1 | 0% | 1,931 | 2,851 | +48% | 0 | 0 | — |
case-04 | pass→pass | 7,236 | 8,096 | +12% | 1 | 1 | 0% | 1,300 | 2,286 | +76% | 0 | 0 | — |
case-05 | pass→fail | 10,315 | 12,721 | +23% | 1 | 1 | 0% | 1,480 | 2,758 | +86% | 0 | 0 | — |
case-06 | pass→pass | 9,347 | 11,539 | +23% | 1 | 1 | 0% | 1,675 | 2,593 | +55% | 0 | 0 | — |
case-07 | pass→pass | 6,844 | 6,604 | -4% | 1 | 1 | 0% | 1,207 | 1,759 | +46% | 0 | 0 | — |
case-08 | pass→pass | 11,100 | 12,151 | +9% | 1 | 1 | 0% | 1,711 | 2,783 | +63% | 0 | 0 | — |
case-09 | pass→pass | 7,549 | 7,736 | +2% | 1 | 1 | 0% | 1,411 | 2,174 | +54% | 0 | 0 | — |
case-10 | pass→pass | 11,826 | 14,456 | +22% | 1 | 1 | 0% | 1,551 | 2,820 | +82% | 0 | 0 | — |
case-11 | pass→pass | 4,550 | 6,318 | +39% | 1 | 1 | 0% | 761 | 1,994 | +162% | 0 | 0 | — |
case-13 | pass→pass | 7,300 | 7,423 | +2% | 1 | 1 | 0% | 1,191 | 2,026 | +70% | 0 | 0 | — |
case-14 | pass→pass | 10,330 | 14,945 | +45% | 1 | 1 | 0% | 1,757 | 2,861 | +63% | 0 | 0 | — |
case-15 | pass→pass | 9,874 | 6,421 | -35% | 1 | 1 | 0% | 1,409 | 1,842 | +31% | 0 | 0 | — |
case-16 | pass→pass | 10,461 | 11,550 | +10% | 1 | 1 | 0% | 1,478 | 2,725 | +84% | 0 | 0 | — |
case-17 | pass→pass | 6,981 | 7,285 | +4% | 1 | 1 | 0% | 1,283 | 2,079 | +62% | 0 | 0 | — |
case-18 | pass→pass | 6,968 | 7,131 | +2% | 1 | 1 | 0% | 1,247 | 2,087 | +67% | 0 | 0 | — |
case-19 | pass→pass | 10,731 | 10,669 | -1% | 1 | 1 | 0% | 1,476 | 2,536 | +72% | 0 | 0 | — |
case-20 | pass→pass | 12,997 | 14,500 | +12% | 1 | 1 | 0% | 1,943 | 3,387 | +74% | 0 | 0 | — |
case-21 | pass→pass | 15,830 | 14,287 | -10% | 1 | 1 | 0% | 2,661 | 2,894 | +9% | 0 | 0 | — |
case-22 | pass→pass | 21,265 | 15,403 | -28% | 1 | 1 | 0% | 2,931 | 3,508 | +20% | 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 0 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.