Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Medusa rules and best practices. These rules should be used when building applications with Medusa.
.claude/skills/amariahak-medusa-rules-and-best-practices/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 27% | 0% |
Medusa rules and best practices. These rules should be used when building applications with Medusa.
Synced from https://github.com/PatrickJS/awesome-cursorrules/tree/main/rules/medusa.mdc.
You are an expert senior software engineer specializing in modern web development, with deep expertise in TypeScript, Medusa, React.js, and TailwindCSS.
MedusaError.@medusajs/framework/workflows-sdk to define it.transform for any data transformation.when to define conditions.await when calling steps.model utility from @medusajs/framework/utils to define data models.model.define should be snake case.id field to a data model, always make it a primary key with .primaryKey().id only, other IDs should be text instead.MedusaService.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 19,722 | 16,255 | -18% | 1 | 1 | 0% | 4,224 | 4,131 | -2% | 0 | 0 | — |
case-02 | fail→fail | 5,649 | 9,283 | +64% | 1 | 1 | 0% | 1,162 | 1,804 | +55% | 0 | 0 | — |
case-03 | fail→pass | 17,517 | 12,813 | -27% | 1 | 1 | 0% | 3,959 | 3,283 | -17% | 0 | 0 | — |
case-04 | pass→pass | 11,686 | 6,006 | -49% | 1 | 1 | 0% | 2,337 | 1,750 | -25% | 0 | 0 | — |
case-05 | pass→pass | 15,621 | 11,639 | -25% | 1 | 1 | 0% | 3,022 | 2,945 | -3% | 0 | 0 | — |
case-06 | pass→pass | 9,440 | 5,828 | -38% | 1 | 1 | 0% | 1,864 | 1,585 | -15% | 0 | 0 | — |
case-07 | pass→pass | 11,559 | 11,477 | -1% | 1 | 1 | 0% | 2,323 | 3,168 | +36% | 0 | 0 | — |
case-08 | fail→pass | 14,933 | 13,847 | -7% | 1 | 1 | 0% | 2,940 | 3,473 | +18% | 0 | 0 | — |
case-09 | pass→pass | 16,944 | 9,879 | -42% | 1 | 1 | 0% | 2,500 | 2,017 | -19% | 0 | 0 | — |
case-10 | pass→pass | 14,192 | 14,015 | -1% | 1 | 1 | 0% | 2,703 | 3,426 | +27% | 0 | 0 | — |
case-11 | pass→pass | 13,017 | 10,639 | -18% | 1 | 1 | 0% | 2,535 | 2,685 | +6% | 0 | 0 | — |
case-12 | pass→pass | 11,232 | 9,678 | -14% | 1 | 1 | 0% | 2,204 | 2,396 | +9% | 0 | 0 | — |
case-13 | pass→pass | 12,609 | 13,830 | +10% | 1 | 1 | 0% | 2,766 | 3,312 | +20% | 0 | 0 | — |
case-14 | pass→pass | 5,281 | 4,914 | -7% | 1 | 1 | 0% | 1,052 | 1,547 | +47% | 0 | 0 | — |
case-15 | pass→pass | 7,419 | 6,179 | -17% | 1 | 1 | 0% | 1,451 | 1,798 | +24% | 0 | 0 | — |
case-16 | fail→pass | 16,314 | 12,494 | -23% | 1 | 1 | 0% | 3,612 | 3,427 | -5% | 0 | 0 | — |
case-17 | fail→pass | 13,621 | 11,121 | -18% | 1 | 1 | 0% | 2,292 | 2,922 | +27% | 0 | 0 | — |
case-18 | fail→pass | 11,366 | 11,529 | +1% | 1 | 1 | 0% | 2,308 | 2,900 | +26% | 0 | 0 | — |
case-19 | pass→pass | 11,141 | 6,732 | -40% | 1 | 1 | 0% | 2,374 | 1,947 | -18% | 0 | 0 | — |
case-20 | pass→pass | 12,560 | 8,179 | -35% | 1 | 1 | 0% | 1,935 | 2,205 | +14% | 0 | 0 | — |
case-21 | pass→pass | 12,813 | 9,439 | -26% | 1 | 1 | 0% | 2,692 | 2,402 | -11% | 0 | 0 | — |
case-22 | pass→pass | 12,550 | 9,321 | -26% | 1 | 1 | 0% | 2,482 | 2,410 | -3% | 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 +27 percentage points is the difference between those two pass rates over the 22 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.