Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →IMMEDIATELY USE THIS SKILL when creating or develop anything and before writing code or implementation plans - refines rough ideas into fully-formed designs through structured Socratic questioning, alternative exploration, and incremental validation
.claude/skills/microsoft-brainstorming/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -54% | 0% |
Transform rough ideas into fully-formed designs through structured questioning and alternative exploration.
Core principle: Ask questions to understand, explore alternatives, present design incrementally for validation.
Announce at start: "I'm using the Nori Brainstorming skill to refine your idea into a design."
When design is approved and implementation will follow:
Ask: "Ready to create the implementation plan?"
When your human partner confirms (any affirmative response):
You can and should go backward when:
Don't force forward linearly when going backward would give better results.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | 7,490 | 2,991 | -60% | 1 | 1 | 0% | 1,250 | 1,067 | -15% | 0 | 0 | — |
case-01 | fail→pass | 6,074 | 3,321 | -45% | 1 | 1 | 0% | 1,074 | 948 | -12% | 0 | 0 | — |
case-02 | fail→pass | 8,549 | 3,184 | -63% | 1 | 1 | 0% | 1,408 | 1,101 | -22% | 0 | 0 | — |
case-03 | fail→pass | 11,603 | 2,283 | -80% | 1 | 1 | 0% | 2,057 | 890 | -57% | 0 | 0 | — |
case-04 | fail→pass | 11,893 | 2,942 | -75% | 1 | 1 | 0% | 2,152 | 998 | -54% | 0 | 0 | — |
case-05 | fail→pass | 15,798 | 2,640 | -83% | 1 | 1 | 0% | 2,740 | 930 | -66% | 0 | 0 | — |
case-06 | fail→pass | 6,850 | 2,493 | -64% | 1 | 1 | 0% | 1,005 | 978 | -3% | 0 | 0 | — |
case-07 | fail→pass | 14,939 | 10,085 | -32% | 1 | 1 | 0% | 2,509 | 2,155 | -14% | 0 | 0 | — |
case-08 | pass→fail | 18,145 | 6,132 | -66% | 1 | 1 | 0% | 3,131 | 1,509 | -52% | 0 | 0 | — |
case-10 | fail→pass | 21,802 | 7,532 | -65% | 1 | 1 | 0% | 4,234 | 1,843 | -56% | 0 | 0 | — |
case-11 | fail→pass | 9,082 | 5,267 | -42% | 1 | 1 | 0% | 1,429 | 1,415 | -1% | 0 | 0 | — |
case-12 | fail→pass | 11,861 | 2,635 | -78% | 1 | 1 | 0% | 1,689 | 999 | -41% | 0 | 0 | — |
case-13 | fail→pass | 7,996 | 1,371 | -83% | 1 | 1 | 0% | 1,318 | 760 | -42% | 0 | 0 | — |
case-14 | fail→pass | 6,425 | 1,971 | -69% | 1 | 1 | 0% | 1,010 | 836 | -17% | 0 | 0 | — |
case-15 | fail→pass | 9,673 | 1,810 | -81% | 1 | 1 | 0% | 1,533 | 849 | -45% | 0 | 0 | — |
case-16 | fail→pass | 12,820 | 6,507 | -49% | 1 | 1 | 0% | 2,094 | 1,662 | -21% | 0 | 0 | — |
case-17 | pass→pass | 8,829 | 4,050 | -54% | 1 | 1 | 0% | 1,541 | 1,271 | -18% | 0 | 0 | — |
case-18 | fail→pass | 12,365 | 2,402 | -81% | 1 | 1 | 0% | 1,875 | 888 | -53% | 0 | 0 | — |
case-19 | pass→pass | 11,227 | 4,833 | -57% | 1 | 1 | 0% | 1,866 | 1,310 | -30% | 0 | 0 | — |
case-20 | pass→pass | 7,209 | 10,218 | +42% | 1 | 1 | 0% | 1,542 | 2,469 | +60% | 0 | 0 | — |
case-21 | fail→fail | 3,175 | 5,464 | +72% | 1 | 1 | 0% | 483 | 1,458 | +202% | 0 | 0 | — |
case-22 | fail→fail | 11,175 | 1,853 | -83% | 1 | 1 | 0% | 1,770 | 832 | -53% | 0 | 0 | — |
case-23 | pass→pass | 13,317 | 7,161 | -46% | 1 | 1 | 0% | 2,262 | 1,748 | -23% | 0 | 0 | — |
case-24 | fail→pass | 14,201 | 2,570 | -82% | 1 | 1 | 0% | 2,266 | 1,017 | -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. 24 cases were attempted. The headline lift of +67 percentage points is the difference between those two pass rates over the 24 comparable cases. 2 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.