Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the user wants to push past conventional workflow limits with advanced performance techniques like parallel orchestration, streaming pipelines, or adaptive routing.
.claude/skills/sharpdeveye-turbocharge/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -3% | 0% |
Invoke /agent-workflow — it contains workflow principles, anti-patterns, and the Context Gathering Protocol. Follow the protocol before proceeding — if no workflow context exists yet, you MUST run /teach-maestro first.
Start your response with:
text──────────── ⚡ TURBOCHARGE ───────────── 》》》 Entering turbocharge mode...
Push a workflow past conventional limits. This isn't about adding features — it's about making existing capabilities operate at a level users didn't think was possible.
EXTRA IMPORTANT: Context determines what "extraordinary" means. Understand the project's scale before deciding what to turbocharge.
Every turbocharge technique must degrade gracefully. The workflow without the enhancement must still work.
After turbocharging, run /evaluate to verify the enhancement works and degrades gracefully.
NEVER:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→pass | 17,163 | 12,403 | -28% | 1 | 1 | 0% | 2,635 | 2,443 | -7% | 0 | 0 | — |
case-06 | fail→fail | 17,192 | 11,335 | -34% | 1 | 1 | 0% | 2,658 | 2,330 | -12% | 0 | 0 | — |
case-07 | fail→pass | 16,610 | 13,326 | -20% | 1 | 1 | 0% | 2,474 | 2,625 | +6% | 0 | 0 | — |
case-01 | pass→fail | 19,875 | 4,338 | -78% | 1 | 1 | 0% | 3,008 | 909 | -70% | 0 | 0 | — |
case-02 | pass→pass | 11,680 | 13,102 | +12% | 1 | 1 | 0% | 1,725 | 2,624 | +52% | 0 | 0 | — |
case-03 | fail→fail | 20,606 | 13,294 | -35% | 1 | 1 | 0% | 3,154 | 2,708 | -14% | 0 | 0 | — |
case-04 | fail→pass | 18,336 | 16,657 | -9% | 1 | 1 | 0% | 2,792 | 3,145 | +13% | 0 | 0 | — |
case-08 | pass→fail | 14,237 | 11,384 | -20% | 1 | 1 | 0% | 2,105 | 2,282 | +8% | 0 | 0 | — |
case-09 | pass→pass | 8,787 | 9,870 | +12% | 1 | 1 | 0% | 1,275 | 2,088 | +64% | 0 | 0 | — |
case-10 | fail→fail | 27,745 | 10,769 | -61% | 1 | 1 | 0% | 2,972 | 2,241 | -25% | 0 | 0 | — |
case-11 | pass→pass | 15,403 | 12,436 | -19% | 1 | 1 | 0% | 2,267 | 2,454 | +8% | 0 | 0 | — |
case-12 | fail→pass | 16,042 | 15,981 | -0% | 1 | 1 | 0% | 2,550 | 2,856 | +12% | 0 | 0 | — |
case-13 | fail→fail | 15,099 | 12,215 | -19% | 1 | 1 | 0% | 2,420 | 2,397 | -1% | 0 | 0 | — |
case-14 | pass→pass | 17,601 | 13,019 | -26% | 1 | 1 | 0% | 2,839 | 2,755 | -3% | 0 | 0 | — |
case-15 | fail→pass | 14,917 | 11,228 | -25% | 1 | 1 | 0% | 2,430 | 2,355 | -3% | 0 | 0 | — |
case-16 | fail→fail | 18,297 | 10,676 | -42% | 1 | 1 | 0% | 2,754 | 2,229 | -19% | 0 | 0 | — |
case-17 | fail→pass | 11,460 | 5,531 | -52% | 1 | 1 | 0% | 1,624 | 1,405 | -13% | 0 | 0 | — |
case-18 | pass→pass | 18,424 | 11,335 | -38% | 1 | 1 | 0% | 2,747 | 2,218 | -19% | 0 | 0 | — |
case-19 | pass→fail | 16,254 | 12,998 | -20% | 1 | 1 | 0% | 2,375 | 2,432 | +2% | 0 | 0 | — |
case-20 | pass→fail | 4,874 | 3,637 | -25% | 1 | 1 | 0% | 786 | 901 | +15% | 0 | 0 | — |
case-21 | pass→fail | 18,320 | 8,581 | -53% | 1 | 1 | 0% | 4,230 | 1,969 | -53% | 0 | 0 | — |
case-22 | fail→fail | 2,344 | 11,223 | +379% | 1 | 1 | 0% | 371 | 2,311 | +523% | 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, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +5 percentage points is the difference between those two pass rates over the 20 comparable cases. 5 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.