Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Query the local Postgres database of the active Aspire worktree via psql.
.claude/skills/platformplatform-db-query/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 133% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 60% | 0% |
Port = .workspace/port.txt base + 4. Never trust Aspire MCP for the port — a common critical failure that silently runs SQL on another worktree's database.
Read-only. Every write needs explicit user approval for that exact statement, every time — prior approvals never carry over.
For destructive operations (DROP, TRUNCATE, DELETE, ALTER, or anything that loses data), take extreme care. If anything in the request is even slightly unclear about the scope, target, or intent, stop and ask for clarification before executing. Always assume the most conservative interpretation.
If psql --version fails, tell the user:
> psql is not installed. Install: > - macOS: brew install libpq && brew link --force libpq > - Linux: sudo apt install postgresql-client > - Windows: winget install PostgreSQL.PostgreSQL.17
Aspire must be running.
localhost, port = <base> + 4 where <base> is the integer in .workspace/port.txt.postgres.export PGPASSWORD=$(dotnet user-secrets list --project application/AppHost/AppHost.csproj | sed -n 's/^postgres-password = //p') (keeps it out of shell history).Each SCS has its own database. List them:
sqlSELECT datname FROM pg_database WHERE datistemplate = false AND datname <> 'postgres';
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | pass→pass | 3,682 | 1,430 | -61% | 1 | 1 | 0% | 650 | 593 | -9% | 0 | 0 | — |
case-11 | fail→fail | 2,488 | 4,068 | +64% | 1 | 1 | 0% | 355 | 770 | +117% | 0 | 0 | — |
case-05 | pass→pass | 7,988 | 3,283 | -59% | 1 | 1 | 0% | 1,340 | 864 | -36% | 0 | 0 | — |
case-01 | fail→fail | 9,875 | 4,876 | -51% | 1 | 1 | 0% | 865 | 563 | -35% | 0 | 0 | — |
case-02 | fail→fail | 5,463 | 4,861 | -11% | 1 | 1 | 0% | 1,013 | 560 | -45% | 0 | 0 | — |
case-03 | fail→fail | 7,269 | 4,022 | -45% | 1 | 1 | 0% | 1,281 | 806 | -37% | 0 | 0 | — |
case-04 | fail→pass | 11,715 | 2,535 | -78% | 1 | 1 | 0% | 1,936 | 823 | -57% | 0 | 0 | — |
case-07 | fail→pass | 7,782 | 1,352 | -83% | 1 | 1 | 0% | 1,268 | 565 | -55% | 0 | 0 | — |
case-08 | fail→pass | 10,351 | 3,534 | -66% | 1 | 1 | 0% | 1,726 | 762 | -56% | 0 | 0 | — |
case-09 | fail→pass | 6,608 | 5,024 | -24% | 1 | 1 | 0% | 509 | 1,185 | +133% | 0 | 0 | — |
case-10 | fail→pass | 4,122 | 7,211 | +75% | 1 | 1 | 0% | 559 | 894 | +60% | 0 | 0 | — |
case-12 | fail→fail | 5,184 | 4,683 | -10% | 1 | 1 | 0% | 766 | 641 | -16% | 0 | 0 | — |
case-13 | pass→pass | 5,257 | 1,492 | -72% | 1 | 1 | 0% | 903 | 522 | -42% | 0 | 0 | — |
case-14 | pass→pass | 8,788 | 3,872 | -56% | 1 | 1 | 0% | 1,548 | 993 | -36% | 0 | 0 | — |
case-15 | fail→pass | 4,642 | 3,220 | -31% | 1 | 1 | 0% | 781 | 810 | +4% | 0 | 0 | — |
case-16 | fail→fail | 7,408 | 7,714 | +4% | 1 | 1 | 0% | 1,206 | 748 | -38% | 0 | 0 | — |
case-17 | fail→pass | 10,381 | 1,494 | -86% | 1 | 1 | 0% | 1,684 | 546 | -68% | 0 | 0 | — |
case-18 | fail→fail | 3,780 | 4,161 | +10% | 1 | 1 | 0% | 408 | 941 | +131% | 0 | 0 | — |
case-19 | pass→fail | 10,337 | 3,959 | -62% | 1 | 1 | 0% | 1,785 | 633 | -65% | 0 | 0 | — |
case-20 | pass→pass | 10,362 | 8,007 | -23% | 1 | 1 | 0% | 1,866 | 1,804 | -3% | 0 | 0 | — |
case-21 | pass→pass | 8,747 | 5,620 | -36% | 1 | 1 | 0% | 1,733 | 1,378 | -20% | 0 | 0 | — |
case-22 | pass→pass | 10,985 | 6,281 | -43% | 1 | 1 | 0% | 2,000 | 1,453 | -27% | 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 18 counted toward the lift figure. The other 4 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 18 comparable cases. 2 cases got worse with the skill loaded, and they are 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.