Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build and analyze three-statement financial models with integrated income statement, balance sheet, and cash flow projections, scenario analysis, and key driver assumptions.
.claude/skills/cowork-os-financial-modeling/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 32% | 0% |
| case-14 | ✓→✗ | ▼ Worse | 67% | 0% |
| case-20 | ✓→✗ | ▼ Worse | 8% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 1% | 0% |
Build and analyze three-statement financial models with integrated income statement, balance sheet, and cash flow projections, scenario analysis, and key driver assumptions.
| Name | Type | Required | Description | |---|---|---|---| | modelType | select | Yes | Type of financial model to build | | company | string | Yes | Company name or ticker symbol | | question | string | Yes | Your specific modeling question | | historicalData | string | No | Historical financial data or key metrics (e.g., last 3 years revenue, margins) | | assumptions | string | No | Key assumptions to use (e.g., 15% revenue growth, 30% EBITDA margin target) |
../financial-modeling.json.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 29,362 | 26,590 | -9% | 1 | 1 | 0% | 6,217 | 6,703 | +8% | 0 | 0 | — |
case-02 | fail→fail | 27,574 | 29,664 | +8% | 1 | 1 | 0% | 6,209 | 6,695 | +8% | 0 | 0 | — |
case-03 | pass→pass | 26,919 | 25,288 | -6% | 1 | 1 | 0% | 6,199 | 6,247 | +1% | 0 | 0 | — |
case-04 | pass→pass | 17,710 | 22,625 | +28% | 1 | 1 | 0% | 3,662 | 5,295 | +45% | 0 | 0 | — |
case-05 | fail→fail | 12,163 | 21,344 | +75% | 1 | 1 | 0% | 1,889 | 2,377 | +26% | 0 | 0 | — |
case-06 | pass→fail | 23,655 | 27,034 | +14% | 1 | 1 | 0% | 5,059 | 6,687 | +32% | 0 | 0 | — |
case-07 | fail→pass | 26,152 | 30,342 | +16% | 1 | 1 | 0% | 6,201 | 6,687 | +8% | 0 | 0 | — |
case-08 | fail→fail | 37,292 | 46,858 | +26% | 1 | 1 | 0% | 6,014 | 6,688 | +11% | 0 | 0 | — |
case-09 | fail→fail | 26,359 | 25,832 | -2% | 1 | 1 | 0% | 6,172 | 6,658 | +8% | 0 | 0 | — |
case-10 | pass→pass | 25,650 | 26,879 | +5% | 1 | 1 | 0% | 6,044 | 6,668 | +10% | 0 | 0 | — |
case-11 | fail→fail | 27,278 | 27,165 | -0% | 1 | 1 | 0% | 6,192 | 6,678 | +8% | 0 | 0 | — |
case-12 | fail→fail | 18,336 | 27,899 | +52% | 1 | 1 | 0% | 3,318 | 6,669 | +101% | 0 | 0 | — |
case-13 | fail→fail | 27,415 | 26,302 | -4% | 1 | 1 | 0% | 6,181 | 6,668 | +8% | 0 | 0 | — |
case-14 | pass→fail | 19,306 | 27,795 | +44% | 1 | 1 | 0% | 4,003 | 6,665 | +67% | 0 | 0 | — |
case-15 | fail→fail | 27,035 | 28,562 | +6% | 1 | 1 | 0% | 5,929 | 6,668 | +12% | 0 | 0 | — |
case-16 | fail→fail | 28,911 | 27,738 | -4% | 1 | 1 | 0% | 6,188 | 6,674 | +8% | 0 | 0 | — |
case-17 | pass→pass | 16,605 | 27,764 | +67% | 1 | 1 | 0% | 3,570 | 6,671 | +87% | 0 | 0 | — |
case-18 | pass→pass | 21,795 | 27,705 | +27% | 1 | 1 | 0% | 5,022 | 6,663 | +33% | 0 | 0 | — |
case-19 | fail→fail | 27,667 | 28,960 | +5% | 1 | 1 | 0% | 6,181 | 6,667 | +8% | 0 | 0 | — |
case-20 | pass→fail | 29,484 | 28,906 | -2% | 1 | 1 | 0% | 6,165 | 6,665 | +8% | 0 | 0 | — |
case-21 | fail→fail | 21,864 | 27,919 | +28% | 1 | 1 | 0% | 5,046 | 6,679 | +32% | 0 | 0 | — |
case-22 | pass→pass | 27,863 | 27,403 | -2% | 1 | 1 | 0% | 6,177 | 6,662 | +8% | 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 -22 percentage points is the difference between those two pass rates over the 22 comparable cases. 5 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.