Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Set up a gratitude practice that survives past week one — a specific format, a realistic cadence, and prompts that avoid the toxic-positivity trap. Use when asked to start a gratitude practice, gratitude journal help, how to be more grateful, or a gratitude routine that sticks. Produces a concrete format (what to write, how many, how specific), an anchor and cadence, variety so it doesn't go stale, and an honest note that gratitude complements — never denies — real difficulty.
.claude/skills/mohitagw15856-gratitude-practice/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 30% | 0% |
Gratitude practices work — but most fizzle because they're vague ("list three things") and quickly become rote, or they curdle into pretending everything's fine. This sets up a specific, sustainable practice: concrete prompts, enough variety to stay fresh, and a grounded stance that holds space for hard things instead of papering over them.
Ask for these if not provided:
Format: # items, how specific, written/spoken]. Anchor: after existing routine], at time]. Cadence: daily / few times a week — why].
Rotate through
Go-deeper option: once a week, one thing in detail — why it mattered.
> Gratitude complements hard feelings; it doesn't replace them. If things are genuinely heavy, this isn't about forcing positivity.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 21,122 | 17,299 | -18% | 1 | 1 | 0% | 2,372 | 2,642 | +11% | 0 | 0 | — |
case-02 | fail→fail | 28,601 | 16,028 | -44% | 1 | 1 | 0% | 3,061 | 2,575 | -16% | 0 | 0 | — |
case-03 | fail→pass | 21,528 | 14,857 | -31% | 1 | 1 | 0% | 2,487 | 2,437 | -2% | 0 | 0 | — |
case-04 | pass→pass | 17,447 | 16,758 | -4% | 1 | 1 | 0% | 1,876 | 2,503 | +33% | 0 | 0 | — |
case-05 | pass→pass | 15,994 | 15,576 | -3% | 1 | 1 | 0% | 1,498 | 2,368 | +58% | 0 | 0 | — |
case-06 | fail→pass | 17,555 | 18,922 | +8% | 1 | 1 | 0% | 1,898 | 2,673 | +41% | 0 | 0 | — |
case-07 | fail→pass | 18,097 | 18,620 | +3% | 1 | 1 | 0% | 1,962 | 2,656 | +35% | 0 | 0 | — |
case-08 | fail→pass | 22,241 | 17,531 | -21% | 1 | 1 | 0% | 2,390 | 2,742 | +15% | 0 | 0 | — |
case-09 | fail→pass | 17,379 | 16,558 | -5% | 1 | 1 | 0% | 2,005 | 2,614 | +30% | 0 | 0 | — |
case-10 | fail→pass | 11,578 | 10,215 | -12% | 1 | 1 | 0% | 1,663 | 2,382 | +43% | 0 | 0 | — |
case-11 | pass→pass | 16,832 | 15,943 | -5% | 1 | 1 | 0% | 1,821 | 2,470 | +36% | 0 | 0 | — |
case-12 | pass→pass | 18,661 | 18,546 | -1% | 1 | 1 | 0% | 2,089 | 2,797 | +34% | 0 | 0 | — |
case-13 | fail→fail | 19,134 | 14,647 | -23% | 1 | 1 | 0% | 2,109 | 2,185 | +4% | 0 | 0 | — |
case-14 | fail→pass | 19,721 | 15,396 | -22% | 1 | 1 | 0% | 2,303 | 2,313 | +0% | 0 | 0 | — |
case-15 | fail→fail | 11,448 | 14,407 | +26% | 1 | 1 | 0% | 1,906 | 2,223 | +17% | 0 | 0 | — |
case-16 | fail→fail | 13,049 | 9,540 | -27% | 1 | 1 | 0% | 1,307 | 2,260 | +73% | 0 | 0 | — |
case-17 | fail→pass | 13,709 | 12,636 | -8% | 1 | 1 | 0% | 2,022 | 2,697 | +33% | 0 | 0 | — |
case-18 | fail→pass | 16,620 | 10,668 | -36% | 1 | 1 | 0% | 1,855 | 2,524 | +36% | 0 | 0 | — |
case-19 | fail→pass | 17,628 | 15,686 | -11% | 1 | 1 | 0% | 1,937 | 2,511 | +30% | 0 | 0 | — |
case-20 | pass→pass | 20,911 | 14,769 | -29% | 1 | 1 | 0% | 2,331 | 3,279 | +41% | 0 | 0 | — |
case-21 | pass→fail | 17,409 | 19,752 | +13% | 1 | 1 | 0% | 2,252 | 3,518 | +56% | 0 | 0 | — |
case-22 | pass→pass | 19,908 | 20,618 | +4% | 1 | 1 | 0% | 2,241 | 3,271 | +46% | 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 +41 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.