Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate scroll-stopping hooks — the first line of a post, thread, video, or email that decides whether anyone keeps reading. Use when asked to write a hook, an opener, a first line, a thread starter, a video cold-open, or to make something more clickable. Produces multiple distinct hook options across proven angles (curiosity, contrarian, result, story, stakes), each labelled with why it works and which platform it fits.
.claude/skills/mohitagw15856-hook-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 81% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 88% | 0% |
The hook is 80% of the result. A brilliant post with a flat first line dies; a mediocre post with a great hook travels. This skill writes hooks the way top creators do — multiple angles, each engineered to stop the scroll — so you can pick the one that fits.
Given just a topic or a finished piece, generate the hooks anyway. Infer the audience and the payoff, and never return a single safe option — the value is in the range. Mark any invented number (assumed — use a real one) because specific numbers are part of what makes hooks land.
Ask for (if not already provided):
Give 8–12 hooks grouped by angle, each with a one-line why it works and the format it suits:
Then:
Keep each hook in the platform's natural length (a YouTube title ≤60 chars; an email subject ≤50; a Reel cold-open speakable in 2–3s).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,048 | 15,295 | +2% | 1 | 1 | 0% | 2,585 | 3,363 | +30% | 0 | 0 | — |
case-02 | fail→pass | 11,936 | 11,336 | -5% | 1 | 1 | 0% | 2,200 | 3,090 | +40% | 0 | 0 | — |
case-03 | fail→fail | 9,760 | 11,615 | +19% | 1 | 1 | 0% | 1,791 | 2,819 | +57% | 0 | 0 | — |
case-04 | fail→fail | 12,210 | 13,544 | +11% | 1 | 1 | 0% | 2,080 | 3,016 | +45% | 0 | 0 | — |
case-05 | fail→pass | 8,996 | 13,047 | +45% | 1 | 1 | 0% | 1,666 | 3,014 | +81% | 0 | 0 | — |
case-06 | fail→fail | 8,283 | 12,311 | +49% | 1 | 1 | 0% | 1,739 | 2,893 | +66% | 0 | 0 | — |
case-07 | fail→fail | 10,136 | 10,483 | +3% | 1 | 1 | 0% | 1,793 | 2,554 | +42% | 0 | 0 | — |
case-08 | fail→fail | 6,688 | 9,909 | +48% | 1 | 1 | 0% | 1,172 | 2,462 | +110% | 0 | 0 | — |
case-09 | fail→fail | 8,219 | 13,042 | +59% | 1 | 1 | 0% | 1,504 | 2,888 | +92% | 0 | 0 | — |
case-10 | fail→fail | 10,284 | 14,018 | +36% | 1 | 1 | 0% | 1,732 | 3,321 | +92% | 0 | 0 | — |
case-11 | fail→fail | 12,241 | 11,098 | -9% | 1 | 1 | 0% | 2,007 | 2,541 | +27% | 0 | 0 | — |
case-12 | fail→fail | 10,126 | 10,536 | +4% | 1 | 1 | 0% | 1,659 | 2,259 | +36% | 0 | 0 | — |
case-13 | pass→fail | 12,323 | 11,322 | -8% | 1 | 1 | 0% | 1,976 | 2,743 | +39% | 0 | 0 | — |
case-14 | fail→pass | 13,577 | 15,917 | +17% | 1 | 1 | 0% | 2,023 | 3,135 | +55% | 0 | 0 | — |
case-15 | fail→pass | 9,605 | 12,121 | +26% | 1 | 1 | 0% | 1,435 | 2,624 | +83% | 0 | 0 | — |
case-16 | fail→fail | 9,967 | 12,608 | +26% | 1 | 1 | 0% | 1,572 | 2,731 | +74% | 0 | 0 | — |
case-17 | fail→pass | 10,632 | 13,997 | +32% | 1 | 1 | 0% | 1,694 | 3,179 | +88% | 0 | 0 | — |
case-18 | fail→pass | 10,245 | 13,660 | +33% | 1 | 1 | 0% | 1,722 | 2,852 | +66% | 0 | 0 | — |
case-19 | fail→pass | 11,497 | 13,805 | +20% | 1 | 1 | 0% | 1,883 | 2,943 | +56% | 0 | 0 | — |
case-20 | pass→pass | 23,275 | 20,725 | -11% | 1 | 1 | 0% | 3,914 | 4,113 | +5% | 0 | 0 | — |
case-21 | pass→fail | 6,270 | 13,763 | +120% | 1 | 1 | 0% | 1,246 | 3,065 | +146% | 0 | 0 | — |
case-22 | pass→fail | 22,209 | 12,499 | -44% | 1 | 1 | 0% | 4,174 | 2,911 | -30% | 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 +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 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.