Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Plan a stargazing session for tonight from where you are — what's worth looking for, when and where to look, and how to see it with just your eyes or basic gear. Use when asked what can I see in the sky tonight, plan stargazing, what's that bright star/planet, or help me find [constellation/planet]. Produces a target list suited to your location, date, and light pollution, a simple when/where-to-look guide, naked-eye vs binocular/telescope notes, and viewing conditions to check — flagging that p
.claude/skills/mohitagw15856-stargazing-tonight/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-21 | ✓→✓ | = Same ✓ | 18% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 5% | 0% |
You don't need a telescope or a dark-sky reserve to have a good night under the stars — you need to know what's up, roughly where to look, and what's realistic from your backyard. This builds a short target list for your location and date, tells you when and which direction to look, and is honest that exact positions shift nightly, so it points you to a live sky map to confirm.
Ask for these if not provided:
Look for
Conditions: Moon brightness x] · check weather/cloud · darker spot: tip]. Beginner tip: how to orient — e.g. find X first, star-hop to Y].
> Positions and times shift nightly and by location — confirm tonight's exact sky in a live planetarium app (many are free).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | pass→pass | 14,971 | 12,260 | -18% | 1 | 1 | 0% | 2,258 | 2,667 | +18% | 0 | 0 | — |
case-01 | fail→pass | 19,784 | 18,126 | -8% | 1 | 1 | 0% | 2,621 | 3,034 | +16% | 0 | 0 | — |
case-02 | pass→pass | 24,219 | 12,177 | -50% | 1 | 1 | 0% | 3,050 | 3,197 | +5% | 0 | 0 | — |
case-03 | pass→pass | 23,478 | 18,747 | -20% | 1 | 1 | 0% | 3,073 | 3,120 | +2% | 0 | 0 | — |
case-04 | pass→pass | 21,352 | 20,234 | -5% | 1 | 1 | 0% | 2,689 | 3,378 | +26% | 0 | 0 | — |
case-05 | pass→pass | 14,345 | 22,042 | +54% | 1 | 1 | 0% | 2,455 | 3,529 | +44% | 0 | 0 | — |
case-06 | pass→pass | 25,256 | 21,194 | -16% | 1 | 1 | 0% | 3,165 | 4,114 | +30% | 0 | 0 | — |
case-07 | pass→pass | 17,008 | 18,240 | +7% | 1 | 1 | 0% | 2,035 | 2,613 | +28% | 0 | 0 | — |
case-08 | pass→pass | 21,310 | 18,437 | -13% | 1 | 1 | 0% | 2,453 | 3,045 | +24% | 0 | 0 | — |
case-09 | fail→fail | 17,999 | 23,117 | +28% | 1 | 1 | 0% | 2,324 | 3,411 | +47% | 0 | 0 | — |
case-10 | fail→pass | 19,487 | 20,007 | +3% | 1 | 1 | 0% | 2,912 | 3,406 | +17% | 0 | 0 | — |
case-11 | pass→pass | 14,832 | 8,347 | -44% | 1 | 1 | 0% | 2,198 | 1,455 | -34% | 0 | 0 | — |
case-12 | pass→pass | 21,847 | 19,785 | -9% | 1 | 1 | 0% | 2,643 | 3,272 | +24% | 0 | 0 | — |
case-13 | pass→pass | 29,434 | 18,523 | -37% | 1 | 1 | 0% | 3,719 | 3,138 | -16% | 0 | 0 | — |
case-14 | pass→pass | 20,810 | 19,113 | -8% | 1 | 1 | 0% | 2,269 | 2,937 | +29% | 0 | 0 | — |
case-15 | fail→pass | 15,861 | 9,230 | -42% | 1 | 1 | 0% | 1,716 | 2,329 | +36% | 0 | 0 | — |
case-16 | pass→pass | 18,594 | 11,328 | -39% | 1 | 1 | 0% | 2,065 | 2,755 | +33% | 0 | 0 | — |
case-17 | pass→pass | 14,034 | 11,343 | -19% | 1 | 1 | 0% | 2,043 | 2,522 | +23% | 0 | 0 | — |
case-18 | pass→pass | 16,694 | 16,930 | +1% | 1 | 1 | 0% | 1,810 | 2,755 | +52% | 0 | 0 | — |
case-19 | pass→pass | 34,421 | 16,584 | -52% | 1 | 1 | 0% | 4,105 | 2,740 | -33% | 0 | 0 | — |
case-20 | pass→pass | 17,272 | 11,937 | -31% | 1 | 1 | 0% | 1,830 | 2,897 | +58% | 0 | 0 | — |
case-22 | pass→pass | 19,798 | 11,655 | -41% | 1 | 1 | 0% | 2,425 | 2,725 | +12% | 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.
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.