Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Public main skill for the incubating optimizer framework. Use when the user wants to analyze performance, identify bottlenecks, design experiments, or validate optimization gains from captures, traces, or profiling evidence. This skill is the future optimizer entry and currently provides the minimum intake contract only.
.claude/skills/haolange-rdc-optimizer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -65% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -15% | 0% |
你是 optimizer framework 的 public main skill 骨架。
当前 framework 仍处于 incubating，所以你的职责只到：
optimizer 而不是 debuggeroptimizer/common/ 作为后续 SSOT 起点至少确认：
optimizer 伪装成已完成的 GA frameworkdebugger/common/ 作为 optimizer 的规则来源| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 9,986 | 13,272 | +33% | 1 | 1 | 0% | 1,799 | 1,618 | -10% | 0 | 0 | — |
case-02 | fail→pass | 21,593 | 11,853 | -45% | 1 | 1 | 0% | 3,313 | 2,009 | -39% | 0 | 0 | — |
case-03 | fail→pass | 12,624 | 8,267 | -35% | 1 | 1 | 0% | 2,223 | 1,540 | -31% | 0 | 0 | — |
case-04 | pass→fail | 18,237 | 6,924 | -62% | 1 | 1 | 0% | 1,735 | 1,266 | -27% | 0 | 0 | — |
case-05 | pass→fail | 24,117 | 9,820 | -59% | 1 | 1 | 0% | 2,451 | 1,830 | -25% | 0 | 0 | — |
case-06 | pass→fail | 18,178 | 12,488 | -31% | 1 | 1 | 0% | 2,940 | 2,360 | -20% | 0 | 0 | — |
case-07 | fail→pass | 27,437 | 8,670 | -68% | 1 | 1 | 0% | 4,028 | 1,426 | -65% | 0 | 0 | — |
case-08 | fail→fail | 13,496 | 7,326 | -46% | 1 | 1 | 0% | 2,263 | 1,240 | -45% | 0 | 0 | — |
case-09 | fail→pass | 18,623 | 16,924 | -9% | 1 | 1 | 0% | 3,230 | 2,732 | -15% | 0 | 0 | — |
case-10 | fail→pass | 16,366 | 9,626 | -41% | 1 | 1 | 0% | 2,636 | 1,750 | -34% | 0 | 0 | — |
case-11 | fail→pass | 18,184 | 10,642 | -41% | 1 | 1 | 0% | 2,956 | 1,900 | -36% | 0 | 0 | — |
case-12 | fail→pass | 23,040 | 13,659 | -41% | 1 | 1 | 0% | 3,798 | 2,496 | -34% | 0 | 0 | — |
case-13 | fail→fail | 16,310 | 8,806 | -46% | 1 | 1 | 0% | 2,853 | 1,619 | -43% | 0 | 0 | — |
case-14 | fail→pass | 15,470 | 6,692 | -57% | 1 | 1 | 0% | 2,798 | 1,514 | -46% | 0 | 0 | — |
case-15 | fail→pass | 25,060 | 10,996 | -56% | 1 | 1 | 0% | 4,534 | 2,027 | -55% | 0 | 0 | — |
case-16 | fail→pass | 13,835 | 9,905 | -28% | 1 | 1 | 0% | 2,331 | 1,670 | -28% | 0 | 0 | — |
case-17 | fail→pass | 21,076 | 7,451 | -65% | 1 | 1 | 0% | 3,932 | 1,352 | -66% | 0 | 0 | — |
case-18 | fail→pass | 17,680 | 11,027 | -38% | 1 | 1 | 0% | 3,386 | 2,062 | -39% | 0 | 0 | — |
case-19 | fail→pass | 45,576 | 12,349 | -73% | 1 | 1 | 0% | 3,707 | 2,211 | -40% | 0 | 0 | — |
case-20 | fail→fail | 19,479 | 12,175 | -37% | 1 | 1 | 0% | 3,340 | 2,212 | -34% | 0 | 0 | — |
case-21 | fail→pass | 18,217 | 12,090 | -34% | 1 | 1 | 0% | 3,150 | 2,093 | -34% | 0 | 0 | — |
case-22 | fail→pass | 23,222 | 17,084 | -26% | 1 | 1 | 0% | 4,404 | 2,911 | -34% | 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 +59 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 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.