Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Validate pending migrations for foreign key consistency, rollback safety, and best practices
.claude/skills/ruvnet-migrate-validate/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 18% | 0% |
Validate all pending database migrations for correctness, safety, and adherence to best practices.
Before applying migrations to catch issues early -- foreign key references to non-existent tables, missing rollback SQL, destructive operations without safeguards, and naming convention violations.
Glob to list all migration files, cross-reference with applied history via mcp__plugin_ruflo-core_ruflo__memory_search --namespace migrations (or memory_list) to identify pending ones. The memory_* tool family routes by namespace; agentdb_hierarchical-* does NOT (it routes by tier), so use memory_* here.Read to load each pending .up.sql and .down.sql file and parse the SQL statementsidx_table_column conventionmcp__plugin_ruflo-core_ruflo__agentdb_pattern-store with type: 'migration-validation'. No namespace arg — ReasoningBank routes it.mcp__plugin_ruflo-core_ruflo__memory_store --namespace migrations for validation results tied to a specific migration number.bashnpx @claude-flow/cli@latest memory search --query "migration validation results" --namespace migrations
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,329 | 3,590 | +8% | 1 | 1 | 0% | 593 | 1,046 | +76% | 0 | 0 | — |
case-02 | fail→fail | 5,599 | 7,274 | +30% | 1 | 1 | 0% | 930 | 670 | -28% | 0 | 0 | — |
case-03 | fail→fail | 2,826 | 5,477 | +94% | 1 | 1 | 0% | 338 | 808 | +139% | 0 | 0 | — |
case-04 | fail→fail | 3,460 | 12,220 | +253% | 1 | 1 | 0% | 614 | 2,686 | +337% | 0 | 0 | — |
case-05 | fail→fail | 7,480 | 8,436 | +13% | 1 | 1 | 0% | 1,475 | 729 | -51% | 0 | 0 | — |
case-06 | fail→fail | 24,901 | 10,850 | -56% | 1 | 1 | 0% | 6,167 | 2,803 | -55% | 0 | 0 | — |
case-07 | fail→pass | 7,435 | 3,326 | -55% | 1 | 1 | 0% | 1,267 | 1,248 | -1% | 0 | 0 | — |
case-08 | pass→pass | 7,490 | 4,421 | -41% | 1 | 1 | 0% | 1,517 | 1,388 | -9% | 0 | 0 | — |
case-09 | pass→pass | 10,978 | 6,623 | -40% | 1 | 1 | 0% | 1,947 | 1,829 | -6% | 0 | 0 | — |
case-10 | pass→fail | 7,844 | 3,583 | -54% | 1 | 1 | 0% | 1,436 | 1,245 | -13% | 0 | 0 | — |
case-11 | pass→pass | 9,822 | 5,257 | -46% | 1 | 1 | 0% | 1,637 | 1,438 | -12% | 0 | 0 | — |
case-12 | pass→pass | 8,837 | 3,899 | -56% | 1 | 1 | 0% | 1,625 | 1,206 | -26% | 0 | 0 | — |
case-13 | pass→pass | 7,304 | 4,095 | -44% | 1 | 1 | 0% | 1,288 | 1,254 | -3% | 0 | 0 | — |
case-14 | fail→pass | 8,209 | 3,817 | -54% | 1 | 1 | 0% | 1,458 | 1,181 | -19% | 0 | 0 | — |
case-15 | fail→pass | 6,032 | 3,106 | -49% | 1 | 1 | 0% | 1,091 | 1,057 | -3% | 0 | 0 | — |
case-16 | fail→pass | 7,403 | 2,291 | -69% | 1 | 1 | 0% | 1,310 | 901 | -31% | 0 | 0 | — |
case-17 | fail→pass | 7,105 | 4,853 | -32% | 1 | 1 | 0% | 1,288 | 1,522 | +18% | 0 | 0 | — |
case-18 | pass→pass | 12,183 | 6,859 | -44% | 1 | 1 | 0% | 2,224 | 1,807 | -19% | 0 | 0 | — |
case-19 | fail→pass | 11,363 | 3,090 | -73% | 1 | 1 | 0% | 1,989 | 919 | -54% | 0 | 0 | — |
case-20 | fail→pass | 9,355 | 5,518 | -41% | 1 | 1 | 0% | 1,656 | 1,516 | -8% | 0 | 0 | — |
case-21 | pass→pass | 7,636 | 6,923 | -9% | 1 | 1 | 0% | 1,294 | 1,319 | +2% | 0 | 0 | — |
case-22 | pass→pass | 2,085 | 2,443 | +17% | 1 | 1 | 0% | 418 | 988 | +136% | 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 19 counted toward the lift figure. The other 3 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 +27 percentage points is the difference between those two pass rates over the 19 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.