Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Strengthen a raw user prompt into an execution-ready instruction set for Amp, Claude Code, Codex, or another AI agent. Use when the user wants to improve an existing prompt, build a reusable prompting framework, wrap the current request with better structure, add clearer tool rules, or create a hook that upgrades prompts before execution.
.claude/skills/reinamaccredy-prompt-leverage/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -71% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -71% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -70% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -81% | 0% |
This skill is maintained in the maestro built-in skills registry. Load the full version:
maestro skill maestro:prompt-leverageThe built-in version includes the framework blocks, provider-specific references (Anthropic Claude, OpenAI GPT), and output mode templates. Loading it from maestro ensures you always get the latest version.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,581 | 17,053 | +9% | 1 | 1 | 0% | 2,580 | 3,014 | +17% | 0 | 0 | — |
case-02 | fail→fail | 15,915 | 13,952 | -12% | 1 | 1 | 0% | 1,979 | 2,325 | +17% | 0 | 0 | — |
case-03 | fail→pass | 16,652 | 3,167 | -81% | 1 | 1 | 0% | 1,637 | 475 | -71% | 0 | 0 | — |
case-04 | fail→pass | 14,014 | 3,292 | -77% | 1 | 1 | 0% | 1,392 | 404 | -71% | 0 | 0 | — |
case-05 | fail→pass | 10,559 | 2,462 | -77% | 1 | 1 | 0% | 1,380 | 417 | -70% | 0 | 0 | — |
case-06 | fail→pass | 14,519 | 5,972 | -59% | 1 | 1 | 0% | 2,256 | 930 | -59% | 0 | 0 | — |
case-07 | fail→pass | 14,742 | 3,441 | -77% | 1 | 1 | 0% | 2,238 | 419 | -81% | 0 | 0 | — |
case-13 | fail→pass | 14,304 | 3,129 | -78% | 1 | 1 | 0% | 2,140 | 380 | -82% | 0 | 0 | — |
case-08 | fail→pass | 15,330 | 3,112 | -80% | 1 | 1 | 0% | 2,654 | 474 | -82% | 0 | 0 | — |
case-09 | fail→pass | 12,341 | 2,333 | -81% | 1 | 1 | 0% | 2,001 | 336 | -83% | 0 | 0 | — |
case-10 | pass→pass | 21,717 | 2,450 | -89% | 1 | 1 | 0% | 2,748 | 437 | -84% | 0 | 0 | — |
case-11 | fail→pass | 24,460 | 1,977 | -92% | 1 | 1 | 0% | 4,387 | 375 | -91% | 0 | 0 | — |
case-12 | fail→pass | 17,369 | 2,236 | -87% | 1 | 1 | 0% | 2,420 | 421 | -83% | 0 | 0 | — |
case-14 | fail→pass | 8,475 | 1,863 | -78% | 1 | 1 | 0% | 1,342 | 267 | -80% | 0 | 0 | — |
case-15 | fail→pass | 11,883 | 6,273 | -47% | 1 | 1 | 0% | 1,891 | 992 | -48% | 0 | 0 | — |
case-16 | fail→pass | 14,920 | 2,625 | -82% | 1 | 1 | 0% | 2,517 | 317 | -87% | 0 | 0 | — |
case-17 | fail→pass | 14,582 | 2,857 | -80% | 1 | 1 | 0% | 2,334 | 563 | -76% | 0 | 0 | — |
case-18 | fail→pass | 14,156 | 2,738 | -81% | 1 | 1 | 0% | 1,830 | 333 | -82% | 0 | 0 | — |
case-19 | fail→pass | 13,538 | 3,077 | -77% | 1 | 1 | 0% | 1,658 | 402 | -76% | 0 | 0 | — |
case-20 | pass→pass | 9,428 | 7,737 | -18% | 1 | 1 | 0% | 1,474 | 1,085 | -26% | 0 | 0 | — |
case-21 | pass→pass | 6,482 | 5,177 | -20% | 1 | 1 | 0% | 1,212 | 1,021 | -16% | 0 | 0 | — |
case-22 | pass→pass | 15,724 | 11,357 | -28% | 1 | 1 | 0% | 2,294 | 1,952 | -15% | 0 | 0 | — |
case-23 | pass→pass | 3,470 | 5,806 | +67% | 1 | 1 | 0% | 532 | 825 | +55% | 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. 23 cases were attempted. The headline lift of +70 percentage points is the difference between those two pass rates over the 23 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.