Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Device controls — vibration, torch, volume, battery status, toast messages. Use for haptic feedback, flashlight, or quick status checks.
.claude/skills/mikeyobrien-device/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | -26% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -36% | 0% |
bashtermux-toast "Message" termux-toast -s "Short message" # shorter duration termux-toast -g top "Top message" # position: top, middle, bottom termux-toast -b black -c white "Styled" # background & text color
bashtermux-vibrate # 1 second default termux-vibrate -d 500 # 500ms termux-vibrate -f # force even in silent mode
bashtermux-torch on termux-torch off
bashtermux-volume # show all streams termux-volume music 10 # set music to 10 termux-volume ring 5 # set ring to 5
Streams: call, system, ring, music, alarm, notification
bashtermux-battery-status
Returns: percentage, status (CHARGING/DISCHARGING), plugged, temperature, etc.
bashtermux-brightness 128 # 0-255 termux-brightness auto # auto brightness
bashtermux-wifi-connectioninfo # current connection termux-wifi-scaninfo # nearby networks termux-wifi-enable true/false # toggle wifi
bashtermux-sensor -l # list available sensors termux-sensor -s accelerometer # read specific sensor termux-sensor -s accelerometer -n 5 # 5 readings then stop
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 9,674 | 8,803 | -9% | 1 | 1 | 0% | 1,506 | 688 | -54% | 0 | 0 | — |
case-02 | fail→pass | 9,325 | 2,419 | -74% | 1 | 1 | 0% | 1,679 | 797 | -53% | 0 | 0 | — |
case-03 | pass→pass | 5,825 | 2,502 | -57% | 1 | 1 | 0% | 1,103 | 815 | -26% | 0 | 0 | — |
case-04 | pass→pass | 7,092 | 2,497 | -65% | 1 | 1 | 0% | 1,258 | 805 | -36% | 0 | 0 | — |
case-05 | pass→pass | 5,692 | 1,785 | -69% | 1 | 1 | 0% | 966 | 694 | -28% | 0 | 0 | — |
case-06 | pass→pass | 5,566 | 1,477 | -73% | 1 | 1 | 0% | 945 | 629 | -33% | 0 | 0 | — |
case-07 | pass→pass | 6,606 | 2,442 | -63% | 1 | 1 | 0% | 1,182 | 749 | -37% | 0 | 0 | — |
case-08 | pass→pass | 8,340 | 2,237 | -73% | 1 | 1 | 0% | 1,475 | 764 | -48% | 0 | 0 | — |
case-09 | pass→pass | 6,624 | 3,066 | -54% | 1 | 1 | 0% | 1,270 | 837 | -34% | 0 | 0 | — |
case-10 | pass→pass | 6,512 | 1,810 | -72% | 1 | 1 | 0% | 1,153 | 657 | -43% | 0 | 0 | — |
case-11 | pass→pass | 6,057 | 1,434 | -76% | 1 | 1 | 0% | 1,102 | 637 | -42% | 0 | 0 | — |
case-12 | pass→pass | 7,637 | 1,920 | -75% | 1 | 1 | 0% | 1,477 | 715 | -52% | 0 | 0 | — |
case-13 | pass→pass | 6,234 | 1,987 | -68% | 1 | 1 | 0% | 1,168 | 717 | -39% | 0 | 0 | — |
case-14 | fail→pass | 8,746 | 2,184 | -75% | 1 | 1 | 0% | 1,565 | 736 | -53% | 0 | 0 | — |
case-15 | pass→pass | 8,275 | 3,264 | -61% | 1 | 1 | 0% | 1,524 | 969 | -36% | 0 | 0 | — |
case-16 | pass→pass | 6,795 | 2,635 | -61% | 1 | 1 | 0% | 1,277 | 833 | -35% | 0 | 0 | — |
case-17 | pass→pass | 9,011 | 6,065 | -33% | 1 | 1 | 0% | 1,657 | 1,457 | -12% | 0 | 0 | — |
case-18 | pass→pass | 9,787 | 3,124 | -68% | 1 | 1 | 0% | 1,702 | 851 | -50% | 0 | 0 | — |
case-19 | pass→pass | 3,455 | 2,455 | -29% | 1 | 1 | 0% | 625 | 779 | +25% | 0 | 0 | — |
case-20 | pass→pass | 8,109 | 6,089 | -25% | 1 | 1 | 0% | 1,510 | 1,179 | -22% | 0 | 0 | — |
case-21 | pass→pass | 5,748 | 2,922 | -49% | 1 | 1 | 0% | 1,018 | 889 | -13% | 0 | 0 | — |
case-22 | pass→pass | 5,689 | 2,136 | -62% | 1 | 1 | 0% | 996 | 732 | -27% | 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.