Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Applies hexagonal architecture isolating domain from infrastructure. Use when designing systems where testability and port/adapter separation are priorities.
.claude/skills/athola-architecture-paradigm-hexagonal/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 50% | 0% |
| case-03 | ✓→✓ | = Same ✓ | -22% | 0% |
These vocabulary items name the concrete tools and abstractions that show up when the paradigm is implemented. They are not required dependencies and they are not part of the skill's tools: frontmatter (which is reserved for Claude Code tool restrictions). Use this list to disambiguate during architecture discussions.
boundary-validator: checks every adapter conforms to the port contractadapter-generator: scaffolds primary and secondary adapters from a port speccontract-tester: validates each adapter against its port via shared contract testsaliases) and stored inside the core domain module, not in adapter directories.
adapter implementations and dependency policies.
directly reference another adapter or the core's internal implementation.
expectations before the adapter is merged.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 20,661 | 18,654 | -10% | 1 | 1 | 0% | 3,150 | 4,175 | +33% | 0 | 0 | — |
case-02 | pass→pass | 22,220 | 26,773 | +20% | 1 | 1 | 0% | 3,713 | 5,562 | +50% | 0 | 0 | — |
case-03 | pass→pass | 14,991 | 7,790 | -48% | 1 | 1 | 0% | 2,637 | 2,068 | -22% | 0 | 0 | — |
case-04 | pass→pass | 10,496 | 5,011 | -52% | 1 | 1 | 0% | 1,497 | 1,544 | +3% | 0 | 0 | — |
case-05 | fail→pass | 36,855 | 6,451 | -82% | 1 | 1 | 0% | 1,265 | 1,900 | +50% | 0 | 0 | — |
case-06 | pass→pass | 15,651 | 11,051 | -29% | 1 | 1 | 0% | 2,632 | 2,707 | +3% | 0 | 0 | — |
case-07 | pass→pass | 10,641 | 4,970 | -53% | 1 | 1 | 0% | 1,766 | 1,685 | -5% | 0 | 0 | — |
case-08 | pass→pass | 16,972 | 14,808 | -13% | 1 | 1 | 0% | 2,456 | 3,172 | +29% | 0 | 0 | — |
case-09 | pass→pass | 15,405 | 10,182 | -34% | 1 | 1 | 0% | 2,493 | 2,398 | -4% | 0 | 0 | — |
case-10 | pass→pass | 12,915 | 6,813 | -47% | 1 | 1 | 0% | 1,895 | 1,924 | +2% | 0 | 0 | — |
case-11 | pass→pass | 12,420 | 10,611 | -15% | 1 | 1 | 0% | 1,987 | 2,536 | +28% | 0 | 0 | — |
case-12 | pass→pass | 15,315 | 12,515 | -18% | 1 | 1 | 0% | 2,289 | 2,766 | +21% | 0 | 0 | — |
case-13 | pass→pass | 13,177 | 10,950 | -17% | 1 | 1 | 0% | 2,161 | 2,679 | +24% | 0 | 0 | — |
case-14 | pass→pass | 13,694 | 10,233 | -25% | 1 | 1 | 0% | 2,085 | 2,357 | +13% | 0 | 0 | — |
case-15 | fail→pass | 10,895 | 2,500 | -77% | 1 | 1 | 0% | 1,646 | 1,286 | -22% | 0 | 0 | — |
case-16 | pass→pass | 11,435 | 3,533 | -69% | 1 | 1 | 0% | 1,644 | 1,431 | -13% | 0 | 0 | — |
case-17 | pass→pass | 14,418 | 10,530 | -27% | 1 | 1 | 0% | 2,181 | 2,524 | +16% | 0 | 0 | — |
case-18 | pass→pass | 15,947 | 14,017 | -12% | 1 | 1 | 0% | 2,341 | 2,899 | +24% | 0 | 0 | — |
case-19 | pass→pass | 16,349 | 13,198 | -19% | 1 | 1 | 0% | 2,562 | 2,930 | +14% | 0 | 0 | — |
case-20 | pass→pass | 19,235 | 12,566 | -35% | 1 | 1 | 0% | 3,243 | 3,013 | -7% | 0 | 0 | — |
case-21 | pass→pass | 18,350 | 19,547 | +7% | 1 | 1 | 0% | 3,093 | 4,115 | +33% | 0 | 0 | — |
case-22 | pass→pass | 8,953 | 1,880 | -79% | 1 | 1 | 0% | 1,345 | 1,178 | -12% | 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 21 counted toward the lift figure. The other 1 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 +14 percentage points is the difference between those two pass rates over the 21 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.