Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build an age-appropriate online-safety plan for a child — the settings, the agreements, and the conversations — that protects without just spying or banning everything. Use when asked to keep my kid safe online, parental controls setup, my child's online safety, or screen rules for kids. Produces an age-tuned plan covering device/platform settings, a family agreement, the ongoing conversations that matter more than any filter, warning signs to watch for, and how to respond to trouble — balancing
.claude/skills/mohitagw15856-kids-online-safety-plan/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 58% | 0% |
Locking everything down breeds workarounds and secrecy; doing nothing leaves kids exposed. Good online safety is settings plus trust: age-appropriate controls, a clear family agreement, and — the part that actually protects — an open line so a kid will come to you when something goes wrong. This builds that, tuned to the child's age.
Ask for these if not provided:
Settings (age-appropriate): device/app/platform controls + privacy defaults]. Family agreement: time · content · sharing · stranger rules — co-created]. Talk about (ongoing): strangers/contact · what's safe to share · pressure & mistakes · coming to you. Warning signs: behavioral changes to watch]. If something goes wrong: calm response · how to report/block · where to get help].
Balance: tighter now / more independence as they show maturity].
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | pass→pass | 23,200 | 20,982 | -10% | 1 | 1 | 0% | 2,397 | 3,179 | +33% | 0 | 0 | — |
case-01 | fail→fail | 25,688 | 28,492 | +11% | 1 | 1 | 0% | 3,413 | 4,446 | +30% | 0 | 0 | — |
case-02 | fail→fail | 29,861 | 29,128 | -2% | 1 | 1 | 0% | 3,381 | 4,001 | +18% | 0 | 0 | — |
case-03 | fail→pass | 27,817 | 24,514 | -12% | 1 | 1 | 0% | 2,997 | 3,509 | +17% | 0 | 0 | — |
case-05 | pass→pass | 15,047 | 15,442 | +3% | 1 | 1 | 0% | 1,414 | 2,344 | +66% | 0 | 0 | — |
case-06 | pass→fail | 20,775 | 21,041 | +1% | 1 | 1 | 0% | 1,988 | 3,142 | +58% | 0 | 0 | — |
case-07 | fail→fail | 30,625 | 25,385 | -17% | 1 | 1 | 0% | 2,428 | 3,494 | +44% | 0 | 0 | — |
case-08 | pass→pass | 18,618 | 22,524 | +21% | 1 | 1 | 0% | 1,756 | 3,113 | +77% | 0 | 0 | — |
case-09 | fail→pass | 22,670 | 23,402 | +3% | 1 | 1 | 0% | 2,379 | 3,610 | +52% | 0 | 0 | — |
case-15 | pass→pass | 21,178 | 27,862 | +32% | 1 | 1 | 0% | 2,412 | 3,859 | +60% | 0 | 0 | — |
case-10 | fail→pass | 24,012 | 9,700 | -60% | 1 | 1 | 0% | 2,508 | 1,589 | -37% | 0 | 0 | — |
case-11 | pass→pass | 21,149 | 24,292 | +15% | 1 | 1 | 0% | 2,385 | 3,401 | +43% | 0 | 0 | — |
case-12 | pass→pass | 24,046 | 24,746 | +3% | 1 | 1 | 0% | 2,549 | 3,298 | +29% | 0 | 0 | — |
case-13 | pass→pass | 20,517 | 22,991 | +12% | 1 | 1 | 0% | 1,937 | 3,189 | +65% | 0 | 0 | — |
case-14 | fail→fail | 21,848 | 20,004 | -8% | 1 | 1 | 0% | 2,077 | 3,012 | +45% | 0 | 0 | — |
case-16 | pass→pass | 32,601 | 26,642 | -18% | 1 | 1 | 0% | 2,456 | 3,710 | +51% | 0 | 0 | — |
case-17 | pass→pass | 26,756 | 53,071 | +98% | 1 | 1 | 0% | 2,330 | 2,880 | +24% | 0 | 0 | — |
case-18 | pass→pass | 26,089 | 22,706 | -13% | 1 | 1 | 0% | 2,937 | 3,873 | +32% | 0 | 0 | — |
case-19 | pass→pass | 17,231 | 28,086 | +63% | 1 | 1 | 0% | 2,315 | 3,930 | +70% | 0 | 0 | — |
case-20 | pass→pass | 22,269 | 26,691 | +20% | 1 | 1 | 0% | 2,183 | 3,559 | +63% | 0 | 0 | — |
case-21 | pass→pass | 28,596 | 23,709 | -17% | 1 | 1 | 0% | 2,409 | 3,484 | +45% | 0 | 0 | — |
case-22 | fail→pass | 74,365 | 24,440 | -67% | 1 | 1 | 0% | 3,309 | 3,736 | +13% | 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. 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.