Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Automatic agent selection and intelligent task routing.
.claude/skills/dokhacgiakhoa-intelligent-routing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 29% | 0% |
Purpose: Automatically analyze user requests and route them to the most appropriate specialist agent(s) without requiring explicit user mentions.
> The AI should act as an intelligent Project Manager, analyzing each request and automatically selecting the best specialist(s) for the job.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 39,984 | 37,917 | -5% | 1 | 1 | 0% | 8,261 | 6,154 | -26% | 0 | 0 | — |
case-02 | fail→pass | 37,436 | 45,955 | +23% | 1 | 1 | 0% | 6,429 | 3,841 | -40% | 0 | 0 | — |
case-03 | fail→fail | 7,953 | 7,986 | +0% | 1 | 1 | 0% | 1,207 | 1,911 | +58% | 0 | 0 | — |
case-04 | fail→fail | 20,462 | 25,237 | +23% | 1 | 1 | 0% | 4,488 | 4,907 | +9% | 0 | 0 | — |
case-05 | fail→fail | 5,743 | 4,877 | -15% | 1 | 1 | 0% | 1,015 | 1,272 | +25% | 0 | 0 | — |
case-06 | fail→pass | 21,665 | 11,674 | -46% | 1 | 1 | 0% | 3,337 | 2,616 | -22% | 0 | 0 | — |
case-07 | pass→pass | 2,959 | 3,590 | +21% | 1 | 1 | 0% | 497 | 1,276 | +157% | 0 | 0 | — |
case-08 | fail→fail | 6,107 | 3,263 | -47% | 1 | 1 | 0% | 687 | 1,229 | +79% | 0 | 0 | — |
case-09 | pass→pass | 15,061 | 14,385 | -4% | 1 | 1 | 0% | 2,652 | 2,667 | +1% | 0 | 0 | — |
case-10 | fail→pass | 13,329 | 15,496 | +16% | 1 | 1 | 0% | 2,275 | 3,518 | +55% | 0 | 0 | — |
case-11 | fail→pass | 30,952 | 13,731 | -56% | 1 | 1 | 0% | 5,861 | 2,983 | -49% | 0 | 0 | — |
case-12 | pass→pass | 3,092 | 5,328 | +72% | 1 | 1 | 0% | 404 | 1,260 | +212% | 0 | 0 | — |
case-13 | pass→pass | 6,201 | 7,021 | +13% | 1 | 1 | 0% | 1,038 | 1,896 | +83% | 0 | 0 | — |
case-14 | fail→pass | 11,971 | 10,536 | -12% | 1 | 1 | 0% | 1,815 | 2,338 | +29% | 0 | 0 | — |
case-15 | fail→fail | 19,498 | 13,567 | -30% | 1 | 1 | 0% | 4,097 | 3,886 | -5% | 0 | 0 | — |
case-16 | pass→pass | 9,829 | 4,731 | -52% | 1 | 1 | 0% | 1,594 | 1,411 | -11% | 0 | 0 | — |
case-17 | pass→pass | 18,447 | 10,406 | -44% | 1 | 1 | 0% | 2,888 | 2,446 | -15% | 0 | 0 | — |
case-18 | fail→pass | 5,693 | 4,896 | -14% | 1 | 1 | 0% | 1,085 | 1,325 | +22% | 0 | 0 | — |
case-19 | fail→pass | 25,137 | 18,683 | -26% | 1 | 1 | 0% | 4,332 | 4,269 | -1% | 0 | 0 | — |
case-20 | pass→pass | 9,956 | 9,874 | -1% | 1 | 1 | 0% | 1,863 | 2,570 | +38% | 0 | 0 | — |
case-21 | pass→pass | 11,912 | 9,172 | -23% | 1 | 1 | 0% | 2,177 | 2,690 | +24% | 0 | 0 | — |
case-22 | pass→pass | 17,746 | 9,854 | -44% | 1 | 1 | 0% | 2,466 | 2,451 | -1% | 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 +32 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.