Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Database design principles and decision-making. Schema design, indexing strategy, ORM selection, serverless databases.
.claude/skills/davila7-database-design/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 19% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | -10% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 25% | 0% |
> Learn to THINK, not copy SQL patterns.
Read ONLY files relevant to the request! Check the content map, find what you need.
| File | Description | When to Read | |------|-------------|--------------| | database-selection.md | PostgreSQL vs Neon vs Turso vs SQLite | Choosing database | | orm-selection.md | Drizzle vs Prisma vs Kysely | Choosing ORM | | schema-design.md | Normalization, PKs, relationships | Designing schema | | indexing.md | Index types, composite indexes | Performance tuning | | optimization.md | N+1, EXPLAIN ANALYZE | Query optimization | | migrations.md | Safe migrations, serverless DBs | Schema changes |
Before designing schema:
❌ Default to PostgreSQL for simple apps (SQLite may suffice) ❌ Skip indexing ❌ Use SELECT in production ❌ Store JSON when structured data is better ❌ Ignore N+1 queries
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→pass | 17,614 | 12,035 | -32% | 1 | 1 | 0% | 3,834 | 2,547 | -34% | 0 | 0 | — |
case-01 | fail→fail | 14,113 | 11,523 | -18% | 1 | 1 | 0% | 2,767 | 2,886 | +4% | 0 | 0 | — |
case-02 | fail→pass | 15,047 | 11,282 | -25% | 1 | 1 | 0% | 2,832 | 2,526 | -11% | 0 | 0 | — |
case-04 | fail→fail | 13,248 | 11,807 | -11% | 1 | 1 | 0% | 2,644 | 2,769 | +5% | 0 | 0 | — |
case-05 | pass→pass | 3,535 | 3,687 | +4% | 1 | 1 | 0% | 797 | 951 | +19% | 0 | 0 | — |
case-06 | pass→pass | 12,517 | 8,480 | -32% | 1 | 1 | 0% | 2,371 | 2,141 | -10% | 0 | 0 | — |
case-07 | pass→pass | 9,862 | 11,042 | +12% | 1 | 1 | 0% | 1,998 | 2,507 | +25% | 0 | 0 | — |
case-08 | pass→pass | 10,413 | 6,994 | -33% | 1 | 1 | 0% | 2,001 | 1,736 | -13% | 0 | 0 | — |
case-09 | pass→pass | 13,483 | 9,207 | -32% | 1 | 1 | 0% | 2,545 | 1,967 | -23% | 0 | 0 | — |
case-10 | pass→pass | 11,148 | 9,769 | -12% | 1 | 1 | 0% | 2,298 | 2,452 | +7% | 0 | 0 | — |
case-11 | fail→fail | 15,500 | 10,585 | -32% | 1 | 1 | 0% | 2,929 | 2,325 | -21% | 0 | 0 | — |
case-12 | fail→fail | 8,947 | 5,965 | -33% | 1 | 1 | 0% | 1,836 | 1,533 | -17% | 0 | 0 | — |
case-13 | pass→pass | 16,940 | 11,182 | -34% | 1 | 1 | 0% | 2,872 | 2,381 | -17% | 0 | 0 | — |
case-14 | pass→pass | 14,401 | 12,007 | -17% | 1 | 1 | 0% | 2,874 | 2,650 | -8% | 0 | 0 | — |
case-15 | pass→pass | 11,647 | 10,984 | -6% | 1 | 1 | 0% | 2,144 | 2,231 | +4% | 0 | 0 | — |
case-16 | pass→pass | 10,240 | 8,099 | -21% | 1 | 1 | 0% | 1,904 | 1,789 | -6% | 0 | 0 | — |
case-17 | pass→pass | 11,943 | 8,367 | -30% | 1 | 1 | 0% | 2,370 | 1,805 | -24% | 0 | 0 | — |
case-18 | pass→pass | 11,353 | 10,505 | -7% | 1 | 1 | 0% | 2,192 | 2,101 | -4% | 0 | 0 | — |
case-19 | pass→pass | 8,702 | 7,967 | -8% | 1 | 1 | 0% | 1,685 | 1,775 | +5% | 0 | 0 | — |
case-20 | pass→pass | 8,434 | 8,421 | -0% | 1 | 1 | 0% | 1,859 | 2,106 | +13% | 0 | 0 | — |
case-21 | pass→pass | 8,159 | 6,378 | -22% | 1 | 1 | 0% | 1,557 | 1,565 | +1% | 0 | 0 | — |
case-22 | pass→pass | 3,379 | 5,466 | +62% | 1 | 1 | 0% | 761 | 1,515 | +99% | 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 +9 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.