Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Iterative consultant agent for building and validating logically consistent 9-block Business Model Canvases.
.claude/skills/sickn33-osterwalder-canvas-architect/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -61% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -43% | 0% |
A specialized architectural tool for designing and auditing business models using Alexander Osterwalder’s 9-block framework. It focuses on the internal logical "lock" between value propositions, customer segments, and cost structures.
The agent iteratively defines the Value Proposition and Customer Segments to ensure they are logically aligned.
The agent builds out the Channels, Relationships, Key Activities, Resources, and Partners.
Final validation to ensure every activity is accounted for in the Cost Structure and revenue streams align with customer segments.
"Draft a Business Model Canvas for an AI-powered agri-tech platform that provides soil analysis for large-scale farmers on a subscription basis. Focus on how the Key Resources (IoT/AI) drive the Cost Structure."
"Analyze the consistency of a direct-to-consumer organic dairy brand. Ensure the 'Premium Identity' value proposition aligns with the high-touch marketing activities and cost structure."
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 20,267 | 15,059 | -26% | 1 | 1 | 0% | 3,484 | 3,137 | -10% | 0 | 0 | — |
case-02 | fail→fail | 18,024 | 14,887 | -17% | 1 | 1 | 0% | 2,630 | 2,952 | +12% | 0 | 0 | — |
case-03 | fail→fail | 20,208 | 16,111 | -20% | 1 | 1 | 0% | 3,367 | 3,173 | -6% | 0 | 0 | — |
case-04 | fail→pass | 25,670 | 8,710 | -66% | 1 | 1 | 0% | 5,776 | 1,888 | -67% | 0 | 0 | — |
case-05 | fail→pass | 19,975 | 5,628 | -72% | 1 | 1 | 0% | 3,494 | 1,374 | -61% | 0 | 0 | — |
case-06 | fail→fail | 23,986 | 18,819 | -22% | 1 | 1 | 0% | 4,022 | 3,816 | -5% | 0 | 0 | — |
case-07 | fail→pass | 16,009 | 11,175 | -30% | 1 | 1 | 0% | 2,581 | 2,305 | -11% | 0 | 0 | — |
case-08 | fail→pass | 18,927 | 10,730 | -43% | 1 | 1 | 0% | 3,220 | 2,167 | -33% | 0 | 0 | — |
case-09 | pass→pass | 14,057 | 12,688 | -10% | 1 | 1 | 0% | 2,400 | 2,442 | +2% | 0 | 0 | — |
case-10 | pass→pass | 12,133 | 7,604 | -37% | 1 | 1 | 0% | 2,070 | 1,753 | -15% | 0 | 0 | — |
case-11 | pass→pass | 13,117 | 10,384 | -21% | 1 | 1 | 0% | 2,239 | 2,184 | -2% | 0 | 0 | — |
case-12 | pass→pass | 12,595 | 7,707 | -39% | 1 | 1 | 0% | 1,941 | 1,775 | -9% | 0 | 0 | — |
case-13 | pass→pass | 17,280 | 7,587 | -56% | 1 | 1 | 0% | 2,725 | 1,679 | -38% | 0 | 0 | — |
case-14 | pass→pass | 12,058 | 11,110 | -8% | 1 | 1 | 0% | 1,921 | 2,205 | +15% | 0 | 0 | — |
case-15 | pass→pass | 10,334 | 8,309 | -20% | 1 | 1 | 0% | 1,747 | 1,733 | -1% | 0 | 0 | — |
case-16 | pass→pass | 17,007 | 9,997 | -41% | 1 | 1 | 0% | 2,694 | 2,002 | -26% | 0 | 0 | — |
case-17 | pass→pass | 17,741 | 13,263 | -25% | 1 | 1 | 0% | 2,808 | 2,642 | -6% | 0 | 0 | — |
case-18 | pass→pass | 14,051 | 9,896 | -30% | 1 | 1 | 0% | 2,214 | 2,047 | -8% | 0 | 0 | — |
case-19 | fail→pass | 19,578 | 8,275 | -58% | 1 | 1 | 0% | 3,025 | 1,729 | -43% | 0 | 0 | — |
case-20 | fail→pass | 16,302 | 8,244 | -49% | 1 | 1 | 0% | 2,574 | 1,782 | -31% | 0 | 0 | — |
case-21 | fail→pass | 12,903 | 9,088 | -30% | 1 | 1 | 0% | 2,152 | 1,999 | -7% | 0 | 0 | — |
case-22 | pass→pass | 16,785 | 10,765 | -36% | 1 | 1 | 0% | 2,486 | 2,212 | -11% | 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. 1 case got worse with the skill loaded, and it is 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.