Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Publish a local Open Design plugin to a new public GitHub repository using gh CLI.
.claude/skills/nexu-io-od-plugin-publish-github/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -84% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -89% | 0% |
Use this workflow when the active project contains a copied plugin folder and the user wants it published as a new public GitHub repository.
plugin_context_path is the copied plugin folder relative to the project working directory.open-design.json, SKILL.md, and any compatibility metadata in the copied folder.curl -sS -X POST "$OD_DAEMON_URL/api/projects/$OD_PROJECT_ID/plugins/publish-github" -H 'content-type: application/json' -d '{"path":"<plugin_context_path>"}'
ok is true, report the final repository URL and any useful log/validation summary.message, code, and useful log lines. When authentication is missing, tell the user to run gh auth login --hostname github.com.Prefer the manifest name as the repository slug. If that repository already exists, choose the next clear slug and mention the rename.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-20 | pass→pass | 18,135 | 6,759 | -63% | 1 | 1 | 0% | 2,500 | 1,497 | -40% | 0 | 0 | — |
case-01 | fail→fail | 6,773 | 8,048 | +19% | 1 | 1 | 0% | 1,017 | 711 | -30% | 0 | 0 | — |
case-02 | fail→fail | 7,484 | 6,444 | -14% | 1 | 1 | 0% | 471 | 634 | +35% | 0 | 0 | — |
case-03 | fail→fail | 8,023 | 4,032 | -50% | 1 | 1 | 0% | 1,173 | 822 | -30% | 0 | 0 | — |
case-04 | pass→pass | 10,997 | 2,107 | -81% | 1 | 1 | 0% | 1,438 | 578 | -60% | 0 | 0 | — |
case-05 | fail→pass | 7,006 | 2,826 | -60% | 1 | 1 | 0% | 1,113 | 667 | -40% | 0 | 0 | — |
case-06 | fail→pass | 8,592 | 1,224 | -86% | 1 | 1 | 0% | 1,314 | 421 | -68% | 0 | 0 | — |
case-07 | pass→pass | 20,118 | 2,367 | -88% | 1 | 1 | 0% | 2,867 | 652 | -77% | 0 | 0 | — |
case-08 | fail→fail | 9,525 | 2,423 | -75% | 1 | 1 | 0% | 1,301 | 645 | -50% | 0 | 0 | — |
case-09 | pass→pass | 9,603 | 2,762 | -71% | 1 | 1 | 0% | 1,426 | 509 | -64% | 0 | 0 | — |
case-10 | pass→pass | 19,058 | 3,304 | -83% | 1 | 1 | 0% | 2,650 | 751 | -72% | 0 | 0 | — |
case-11 | fail→fail | 19,415 | 1,811 | -91% | 1 | 1 | 0% | 3,136 | 484 | -85% | 0 | 0 | — |
case-12 | fail→pass | 12,531 | 3,146 | -75% | 1 | 1 | 0% | 1,799 | 801 | -55% | 0 | 0 | — |
case-13 | fail→pass | 19,708 | 2,089 | -89% | 1 | 1 | 0% | 3,671 | 600 | -84% | 0 | 0 | — |
case-14 | fail→pass | 24,652 | 2,385 | -90% | 1 | 1 | 0% | 4,336 | 498 | -89% | 0 | 0 | — |
case-15 | pass→pass | 9,610 | 3,929 | -59% | 1 | 1 | 0% | 1,341 | 828 | -38% | 0 | 0 | — |
case-16 | fail→fail | 11,548 | 6,123 | -47% | 1 | 1 | 0% | 1,886 | 638 | -66% | 0 | 0 | — |
case-17 | fail→pass | 10,797 | 4,266 | -60% | 1 | 1 | 0% | 1,864 | 924 | -50% | 0 | 0 | — |
case-18 | pass→pass | 25,491 | 2,134 | -92% | 1 | 1 | 0% | 3,983 | 582 | -85% | 0 | 0 | — |
case-19 | pass→fail | 12,549 | 3,608 | -71% | 1 | 1 | 0% | 1,865 | 813 | -56% | 0 | 0 | — |
case-21 | pass→pass | 9,070 | 8,063 | -11% | 1 | 1 | 0% | 1,549 | 1,667 | +8% | 0 | 0 | — |
case-22 | pass→pass | 9,530 | 6,177 | -35% | 1 | 1 | 0% | 1,505 | 1,260 | -16% | 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 20 counted toward the lift figure. The other 2 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 20 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.