Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Help you write a heartfelt letter to someone you love — for an anniversary, a hard time, a birthday, or just because — that sounds like you and says what you actually mean. Use when asked to help me write a love letter, say how I feel to my partner, a heartfelt note for [occasion], or I'm not good with words. Produces a letter built from your real feelings and specifics, a structure that carries emotion without cheese, the right tone for your relationship and occasion, and phrasing in your own v
.claude/skills/mohitagw15856-love-letter-helper/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 48% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 104% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 44% | 0% |
Plenty of people feel a lot and freeze at the blank page. This helps you say it — drawing out the specific things you love and want to express, then shaping them into a letter that sounds like you, not a greeting card. It works for an anniversary, an apology-adjacent "I appreciate you," a rough patch, or no reason at all.
Ask for these if not provided:
If you're stuck, tell me: 2–3 prompts to draw out specifics].
Your letter (draft) > Natural opening] … the heart — specific moments/qualities/what they mean, said sincerely] … warm close].
In your voice: note on keeping it sounding like them]. Adjust: make it shorter/warmer/more playful — easy dials].
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,289 | 14,320 | -6% | 1 | 1 | 0% | 1,788 | 2,467 | +38% | 0 | 0 | — |
case-02 | fail→pass | 13,282 | 7,794 | -41% | 1 | 1 | 0% | 1,381 | 2,169 | +57% | 0 | 0 | — |
case-03 | fail→pass | 14,186 | 13,604 | -4% | 1 | 1 | 0% | 1,462 | 2,166 | +48% | 0 | 0 | — |
case-04 | fail→fail | 14,633 | 11,159 | -24% | 1 | 1 | 0% | 2,258 | 2,597 | +15% | 0 | 0 | — |
case-05 | pass→fail | 12,747 | 16,805 | +32% | 1 | 1 | 0% | 1,328 | 2,797 | +111% | 0 | 0 | — |
case-06 | pass→pass | 14,577 | 19,188 | +32% | 1 | 1 | 0% | 1,738 | 4,272 | +146% | 0 | 0 | — |
case-07 | pass→pass | 14,128 | 16,716 | +18% | 1 | 1 | 0% | 2,128 | 2,813 | +32% | 0 | 0 | — |
case-08 | fail→fail | 7,746 | 7,222 | -7% | 1 | 1 | 0% | 1,261 | 2,077 | +65% | 0 | 0 | — |
case-09 | fail→fail | 10,003 | 16,343 | +63% | 1 | 1 | 0% | 1,525 | 2,570 | +69% | 0 | 0 | — |
case-10 | fail→pass | 5,929 | 11,313 | +91% | 1 | 1 | 0% | 948 | 1,932 | +104% | 0 | 0 | — |
case-11 | fail→fail | 14,645 | 11,353 | -22% | 1 | 1 | 0% | 1,534 | 2,599 | +69% | 0 | 0 | — |
case-12 | fail→fail | 15,941 | 15,516 | -3% | 1 | 1 | 0% | 1,713 | 2,348 | +37% | 0 | 0 | — |
case-13 | pass→pass | 8,291 | 10,687 | +29% | 1 | 1 | 0% | 532 | 1,714 | +222% | 0 | 0 | — |
case-14 | pass→pass | 10,852 | 13,604 | +25% | 1 | 1 | 0% | 1,681 | 2,209 | +31% | 0 | 0 | — |
case-21 | pass→pass | 10,973 | 13,605 | +24% | 1 | 1 | 0% | 945 | 2,253 | +138% | 0 | 0 | — |
case-15 | fail→pass | 14,703 | 9,087 | -38% | 1 | 1 | 0% | 1,579 | 2,274 | +44% | 0 | 0 | — |
case-16 | fail→fail | 9,651 | 7,920 | -18% | 1 | 1 | 0% | 1,584 | 2,154 | +36% | 0 | 0 | — |
case-17 | fail→pass | 13,687 | 13,485 | -1% | 1 | 1 | 0% | 1,478 | 2,166 | +47% | 0 | 0 | — |
case-18 | pass→pass | 12,578 | 14,201 | +13% | 1 | 1 | 0% | 1,709 | 2,216 | +30% | 0 | 0 | — |
case-19 | fail→pass | 11,049 | 13,005 | +18% | 1 | 1 | 0% | 1,765 | 2,193 | +24% | 0 | 0 | — |
case-20 | fail→pass | 14,958 | 14,001 | -6% | 1 | 1 | 0% | 1,460 | 2,323 | +59% | 0 | 0 | — |
case-22 | pass→pass | 9,881 | 9,173 | -7% | 1 | 1 | 0% | 1,366 | 2,104 | +54% | 0 | 0 | — |
case-23 | fail→pass | 10,431 | 8,540 | -18% | 1 | 1 | 0% | 1,572 | 2,077 | +32% | 0 | 0 | — |
case-24 | fail→fail | 10,368 | 11,860 | +14% | 1 | 1 | 0% | 1,467 | 1,917 | +31% | 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. 24 cases were attempted. The headline lift of +33 percentage points is the difference between those two pass rates over the 24 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.