Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Make a repository Superset-ready — author .superset/config.json with setup/teardown/run scripts so every new workspace boots configured, then verify with a real workspace. Use when the user wants to set up a project or repo for Superset, configure workspace setup scripts, or fix a failing workspace setup.
.claude/skills/superset-sh-setup/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 56% | 0% |
Goal: every new workspace (isolated git worktree) for this repo comes up ready — dependencies installed, env present, services reachable — without manual steps.
Work out what a fresh worktree needs, and ask about anything ambiguous:
.env is usually gitignored, so new worktrees need it copied from the main checkout or generated from .env.examplecwd?).superset/config.jsonThis file is what wires lifecycle scripts in — a bare setup.sh without config.json is NOT picked up. Schema:
json{ "setup": ["./.superset/setup.sh"], "teardown": ["./.superset/teardown.sh"], "run": ["bun dev"], "cwd": "optional/subdir" }
Each key is an array of shell commands run inside the worktree on workspace create / delete / run. Guidelines:
.superset/config.local.json (gitignored) lets an individual user extend scripts with before/after arrays without touching the shared configShow the user the proposed files and get explicit approval before writing.
Create a throwaway workspace with superset workspaces create --project <id> --name "setup-test" and watch the "Workspace Setup" terminal output. Fix and repeat until it completes cleanly, then delete the test workspace. Setup is not done until a real workspace boots green.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | pass→pass | 17,773 | 15,281 | -14% | 1 | 1 | 0% | 2,827 | 2,963 | +5% | 0 | 0 | — |
case-22 | pass→pass | 10,939 | 17,477 | +60% | 1 | 1 | 0% | 1,976 | 2,514 | +27% | 0 | 0 | — |
case-07 | pass→pass | 16,289 | 9,995 | -39% | 1 | 1 | 0% | 2,811 | 2,237 | -20% | 0 | 0 | — |
case-05 | fail→pass | 8,686 | 1,711 | -80% | 1 | 1 | 0% | 1,516 | 746 | -51% | 0 | 0 | — |
case-01 | fail→fail | 3,942 | 3,902 | -1% | 1 | 1 | 0% | 165 | 580 | +252% | 0 | 0 | — |
case-02 | fail→fail | 5,870 | 3,005 | -49% | 1 | 1 | 0% | 437 | 781 | +79% | 0 | 0 | — |
case-03 | fail→fail | 2,746 | 2,971 | +8% | 1 | 1 | 0% | 196 | 704 | +259% | 0 | 0 | — |
case-04 | pass→pass | 13,850 | 3,837 | -72% | 1 | 1 | 0% | 2,166 | 1,060 | -51% | 0 | 0 | — |
case-08 | pass→pass | 10,631 | 6,574 | -38% | 1 | 1 | 0% | 1,662 | 1,570 | -6% | 0 | 0 | — |
case-09 | fail→pass | 26,215 | 4,051 | -85% | 1 | 1 | 0% | 1,115 | 1,089 | -2% | 0 | 0 | — |
case-10 | pass→pass | 9,109 | 5,226 | -43% | 1 | 1 | 0% | 1,622 | 1,370 | -16% | 0 | 0 | — |
case-11 | fail→pass | 10,390 | 4,659 | -55% | 1 | 1 | 0% | 1,750 | 1,260 | -28% | 0 | 0 | — |
case-12 | pass→pass | 2,621 | 2,233 | -15% | 1 | 1 | 0% | 430 | 866 | +101% | 0 | 0 | — |
case-13 | pass→pass | 4,234 | 4,673 | +10% | 1 | 1 | 0% | 697 | 1,210 | +74% | 0 | 0 | — |
case-14 | pass→pass | 14,489 | 7,834 | -46% | 1 | 1 | 0% | 2,148 | 1,618 | -25% | 0 | 0 | — |
case-15 | pass→pass | 10,137 | 2,112 | -79% | 1 | 1 | 0% | 1,750 | 807 | -54% | 0 | 0 | — |
case-16 | fail→pass | 11,537 | 1,752 | -85% | 1 | 1 | 0% | 1,948 | 672 | -66% | 0 | 0 | — |
case-17 | pass→pass | 9,443 | 4,106 | -57% | 1 | 1 | 0% | 1,494 | 1,082 | -28% | 0 | 0 | — |
case-18 | pass→pass | 7,929 | 5,768 | -27% | 1 | 1 | 0% | 1,168 | 1,218 | +4% | 0 | 0 | — |
case-19 | fail→pass | 4,332 | 3,692 | -15% | 1 | 1 | 0% | 682 | 1,067 | +56% | 0 | 0 | — |
case-20 | pass→pass | 16,875 | 16,549 | -2% | 1 | 1 | 0% | 3,083 | 2,986 | -3% | 0 | 0 | — |
case-21 | pass→pass | 9,845 | 10,676 | +8% | 1 | 1 | 0% | 1,834 | 2,370 | +29% | 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 +23 percentage points is the difference between those two pass rates over the 18 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.