Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Route freeform text to the right GSD command automatically
.claude/skills/coco-research-gsd-do/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 1040% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 252% | 0% |
<objective> Analyze freeform natural language input and dispatch to the most appropriate GSD command.
Acts as a smart dispatcher — never does the work itself. Matches intent to the best GSD command using routing rules, confirms the match, then hands off.
Use when you know what you want but don't know which /gsd-* command to run. </objective>
<execution_context> @$HOME/.claude/get-shit-done/workflows/do.md @$HOME/.claude/get-shit-done/references/ui-brand.md </execution_context>
<context> $ARGUMENTS </context>
<process> Execute the do workflow from @$HOME/.claude/get-shit-done/workflows/do.md end-to-end. Route user intent to the best GSD command and invoke it. </process>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 12,681 | 11,980 | -6% | 1 | 1 | 0% | 1,287 | 1,397 | +9% | 0 | 0 | — |
case-02 | pass→pass | 10,308 | 11,221 | +9% | 1 | 1 | 0% | 608 | 1,137 | +87% | 0 | 0 | — |
case-03 | fail→pass | 11,293 | 7,516 | -33% | 1 | 1 | 0% | 980 | 1,363 | +39% | 0 | 0 | — |
case-04 | fail→fail | 14,798 | 15,849 | +7% | 1 | 1 | 0% | 1,468 | 469 | -68% | 0 | 0 | — |
case-05 | fail→pass | 7,559 | 22,692 | +200% | 1 | 1 | 0% | 282 | 3,215 | +1040% | 0 | 0 | — |
case-06 | fail→pass | 7,673 | 6,207 | -19% | 1 | 1 | 0% | 1,438 | 1,189 | -17% | 0 | 0 | — |
case-07 | pass→fail | 11,266 | 11,564 | +3% | 1 | 1 | 0% | 1,007 | 586 | -42% | 0 | 0 | — |
case-08 | fail→pass | 7,639 | 10,715 | +40% | 1 | 1 | 0% | 1,034 | 1,220 | +18% | 0 | 0 | — |
case-09 | pass→pass | 10,359 | 7,614 | -26% | 1 | 1 | 0% | 860 | 1,499 | +74% | 0 | 0 | — |
case-10 | fail→fail | 16,644 | 13,348 | -20% | 1 | 1 | 0% | 2,919 | 968 | -67% | 0 | 0 | — |
case-11 | fail→pass | 2,704 | 18,199 | +573% | 1 | 1 | 0% | 464 | 1,634 | +252% | 0 | 0 | — |
case-12 | fail→pass | 9,533 | 7,699 | -19% | 1 | 1 | 0% | 1,313 | 1,461 | +11% | 0 | 0 | — |
case-13 | fail→pass | 20,983 | 7,875 | -62% | 1 | 1 | 0% | 1,221 | 1,354 | +11% | 0 | 0 | — |
case-14 | fail→pass | 13,908 | 21,731 | +56% | 1 | 1 | 0% | 1,105 | 1,548 | +40% | 0 | 0 | — |
case-15 | fail→pass | 16,326 | 8,533 | -48% | 1 | 1 | 0% | 1,172 | 1,617 | +38% | 0 | 0 | — |
case-16 | fail→pass | 14,131 | 13,681 | -3% | 1 | 1 | 0% | 1,167 | 1,335 | +14% | 0 | 0 | — |
case-17 | fail→pass | 19,662 | 11,245 | -43% | 1 | 1 | 0% | 567 | 2,040 | +260% | 0 | 0 | — |
case-18 | fail→pass | 14,809 | 10,267 | -31% | 1 | 1 | 0% | 1,429 | 1,708 | +20% | 0 | 0 | — |
case-19 | fail→pass | 13,923 | 12,215 | -12% | 1 | 1 | 0% | 1,592 | 1,201 | -25% | 0 | 0 | — |
case-20 | pass→pass | 8,580 | 14,170 | +65% | 1 | 1 | 0% | 509 | 1,842 | +262% | 0 | 0 | — |
case-21 | pass→pass | 16,622 | 11,422 | -31% | 1 | 1 | 0% | 1,935 | 2,080 | +7% | 0 | 0 | — |
case-22 | pass→pass | 24,003 | 19,325 | -19% | 1 | 1 | 0% | 3,135 | 3,101 | -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, and 18 counted toward the lift figure. The other 4 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 +55 percentage points is the difference between those two pass rates over the 18 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.