Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Start or restart the .NET Aspire AppHost via the developer CLI. Always use this, never the developer CLI's `run` command, `aspire run`, or `aspire restart`.
.claude/skills/platformplatform-aspire-restart/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -48% | 0% |
bashdotnet run --project developer-cli -- restart
Use developer-cli exactly as written - do not expand to an absolute worktree path.
Stops any running Aspire AppHost and starts a fresh instance. Detached by default - the CLI returns immediately while Aspire keeps running.
Always use restart, even when nothing is running yet. It is a no-op when Aspire is not up, and the safe default in every other case. Never use the developer CLI's run command, aspire run, or aspire restart.
--public-url <url> - set PUBLIC_URL (e.g. an ngrok URL)restart is the safe default; it works whether Aspire is up or not.Fully automatic. On a fresh checkout the developer CLI bootstraps .workspace/port.txt: the root gets the default base port; worktrees scan a fixed list of candidates and pick the first one whose ports are all free. Once written, the file is the authoritative allocation for the lifetime of the checkout.
The CLI prints Aspire is restarting on https://app.dev.localhost:<port> and exits before Aspire is fully ready.
To stop without restarting, run dotnet run --project developer-cli -- stop. Rarely needed - prefer restart for everyday use.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,289 | 1,636 | -69% | 1 | 1 | 0% | 782 | 522 | -33% | 0 | 0 | — |
case-02 | fail→pass | 7,828 | 2,190 | -72% | 1 | 1 | 0% | 1,358 | 563 | -59% | 0 | 0 | — |
case-03 | fail→fail | 4,746 | 3,156 | -34% | 1 | 1 | 0% | 741 | 479 | -35% | 0 | 0 | — |
case-04 | fail→pass | 5,644 | 2,242 | -60% | 1 | 1 | 0% | 1,026 | 734 | -28% | 0 | 0 | — |
case-05 | fail→fail | 6,154 | 1,951 | -68% | 1 | 1 | 0% | 972 | 610 | -37% | 0 | 0 | — |
case-06 | fail→pass | 10,221 | 2,869 | -72% | 1 | 1 | 0% | 1,678 | 803 | -52% | 0 | 0 | — |
case-07 | fail→fail | 8,376 | 1,714 | -80% | 1 | 1 | 0% | 1,276 | 585 | -54% | 0 | 0 | — |
case-08 | pass→pass | 8,578 | 2,524 | -71% | 1 | 1 | 0% | 1,382 | 733 | -47% | 0 | 0 | — |
case-09 | fail→pass | 9,763 | 1,932 | -80% | 1 | 1 | 0% | 1,734 | 630 | -64% | 0 | 0 | — |
case-10 | fail→pass | 10,429 | 3,337 | -68% | 1 | 1 | 0% | 1,770 | 919 | -48% | 0 | 0 | — |
case-11 | fail→pass | 11,311 | 3,265 | -71% | 1 | 1 | 0% | 1,913 | 895 | -53% | 0 | 0 | — |
case-12 | fail→pass | 8,234 | 2,696 | -67% | 1 | 1 | 0% | 1,293 | 773 | -40% | 0 | 0 | — |
case-18 | pass→pass | 26,827 | 1,285 | -95% | 1 | 1 | 0% | 5,033 | 550 | -89% | 0 | 0 | — |
case-13 | fail→pass | 9,797 | 1,832 | -81% | 1 | 1 | 0% | 1,684 | 628 | -63% | 0 | 0 | — |
case-14 | fail→fail | 12,678 | 1,621 | -87% | 1 | 1 | 0% | 2,076 | 625 | -70% | 0 | 0 | — |
case-15 | fail→pass | 10,076 | 2,932 | -71% | 1 | 1 | 0% | 1,740 | 784 | -55% | 0 | 0 | — |
case-16 | fail→pass | 10,990 | 2,356 | -79% | 1 | 1 | 0% | 1,805 | 696 | -61% | 0 | 0 | — |
case-17 | pass→pass | 10,309 | 1,775 | -83% | 1 | 1 | 0% | 1,509 | 655 | -57% | 0 | 0 | — |
case-19 | pass→fail | 3,426 | 3,776 | +10% | 1 | 1 | 0% | 531 | 951 | +79% | 0 | 0 | — |
case-20 | pass→pass | 5,987 | 4,190 | -30% | 1 | 1 | 0% | 988 | 1,031 | +4% | 0 | 0 | — |
case-21 | pass→pass | 8,294 | 4,043 | -51% | 1 | 1 | 0% | 1,522 | 1,075 | -29% | 0 | 0 | — |
case-22 | pass→pass | 4,481 | 3,070 | -31% | 1 | 1 | 0% | 742 | 801 | +8% | 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 +41 percentage points is the difference between those two pass rates over the 22 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.