Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Provides quick acknowledgment during voice latency gaps. Use in voice mode o ruser says quick ack, thinking aloud, fast think. Use when the user needs th iscapability. Triggers: quick ack, thinking aloud, fast think, voice ack.
.claude/skills/stijnman-voice-think-fast-handler/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -36% | 0% |
real-time-voice-reasonervoice-synthesis-handler| Failure | Response | |---------|----------| | Double ack | Suppress duplicate acks in same turn. |
Input: User request matching triggers above. Output: Structured result per workflow with integrations invoked as needed.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 10,259 | 23,684 | +131% | 1 | 1 | 0% | 2,105 | 1,818 | -14% | 0 | 0 | — |
case-20 | fail→fail | 15,154 | 10,630 | -30% | 1 | 1 | 0% | 2,131 | 2,084 | -2% | 0 | 0 | — |
case-02 | fail→fail | 2,396 | 2,975 | +24% | 1 | 1 | 0% | 352 | 560 | +59% | 0 | 0 | — |
case-03 | pass→pass | 7,768 | 4,091 | -47% | 1 | 1 | 0% | 946 | 837 | -12% | 0 | 0 | — |
case-04 | pass→pass | 4,921 | 3,801 | -23% | 1 | 1 | 0% | 802 | 709 | -12% | 0 | 0 | — |
case-05 | pass→pass | 5,821 | 5,043 | -13% | 1 | 1 | 0% | 927 | 875 | -6% | 0 | 0 | — |
case-10 | pass→pass | 7,325 | 7,220 | -1% | 1 | 1 | 0% | 1,100 | 1,147 | +4% | 0 | 0 | — |
case-06 | pass→pass | 7,758 | 7,950 | +2% | 1 | 1 | 0% | 1,348 | 1,495 | +11% | 0 | 0 | — |
case-07 | fail→pass | 15,202 | 3,804 | -75% | 1 | 1 | 0% | 2,129 | 890 | -58% | 0 | 0 | — |
case-08 | fail→pass | 10,262 | 3,050 | -70% | 1 | 1 | 0% | 1,754 | 570 | -68% | 0 | 0 | — |
case-09 | fail→pass | 7,088 | 3,209 | -55% | 1 | 1 | 0% | 1,181 | 541 | -54% | 0 | 0 | — |
case-11 | pass→pass | 3,990 | 3,705 | -7% | 1 | 1 | 0% | 413 | 743 | +80% | 0 | 0 | — |
case-12 | fail→fail | 3,724 | 4,704 | +26% | 1 | 1 | 0% | 578 | 831 | +44% | 0 | 0 | — |
case-13 | pass→pass | 13,705 | 4,040 | -71% | 1 | 1 | 0% | 2,304 | 829 | -64% | 0 | 0 | — |
case-14 | pass→pass | 17,024 | 3,751 | -78% | 1 | 1 | 0% | 2,217 | 839 | -62% | 0 | 0 | — |
case-15 | fail→fail | 13,493 | 3,071 | -77% | 1 | 1 | 0% | 1,552 | 694 | -55% | 0 | 0 | — |
case-16 | fail→pass | 4,939 | 2,689 | -46% | 1 | 1 | 0% | 904 | 634 | -30% | 0 | 0 | — |
case-17 | pass→pass | 5,562 | 4,615 | -17% | 1 | 1 | 0% | 942 | 1,096 | +16% | 0 | 0 | — |
case-18 | fail→pass | 11,968 | 6,422 | -46% | 1 | 1 | 0% | 1,649 | 1,051 | -36% | 0 | 0 | — |
case-19 | pass→fail | 6,921 | 3,003 | -57% | 1 | 1 | 0% | 1,039 | 606 | -42% | 0 | 0 | — |
case-21 | fail→fail | 8,696 | 10,750 | +24% | 1 | 1 | 0% | 1,487 | 1,654 | +11% | 0 | 0 | — |
case-22 | pass→fail | 9,675 | 6,852 | -29% | 1 | 1 | 0% | 1,669 | 1,379 | -17% | 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. 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.