Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A simple "Hello World" skill to test your OMA installation
.claude/skills/bilal140202-hello-antigravity/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -41% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -63% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -10% | 0% |
This is a demonstration skill for Oh My Antigravity. It shows the basic structure of a skill (plugin) that Antigravity can understand.
When you need a friendly greeting or want to test if OMA is working correctly, just mention "hello antigravity" or ask me to use this skill.
🎉 Hello from Oh My Antigravity!
Your OMA installation is working perfectly.
This skill was loaded from: ~/.gemini/antigravity/skills/hello-antigravity/
Keep building amazing things!To create a new skill:
plugins/ (e.g., plugins/my-skill/)SKILL.md file with YAML frontmatterscripts/, examples/, or resources/oma install my-skillThat's it! Antigravity will now be able to use your custom skill.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,879 | 2,388 | -59% | 1 | 1 | 0% | 1,030 | 612 | -41% | 0 | 0 | — |
case-02 | fail→pass | 4,352 | 1,679 | -61% | 1 | 1 | 0% | 721 | 573 | -21% | 0 | 0 | — |
case-03 | fail→pass | 3,914 | 2,593 | -34% | 1 | 1 | 0% | 600 | 725 | +21% | 0 | 0 | — |
case-04 | pass→pass | 7,998 | 2,593 | -68% | 1 | 1 | 0% | 1,498 | 836 | -44% | 0 | 0 | — |
case-05 | pass→pass | 9,491 | 2,445 | -74% | 1 | 1 | 0% | 1,731 | 762 | -56% | 0 | 0 | — |
case-06 | pass→pass | 7,872 | 3,082 | -61% | 1 | 1 | 0% | 1,406 | 868 | -38% | 0 | 0 | — |
case-07 | fail→pass | 5,343 | 1,771 | -67% | 1 | 1 | 0% | 1,013 | 595 | -41% | 0 | 0 | — |
case-08 | pass→pass | 7,048 | 2,580 | -63% | 1 | 1 | 0% | 1,168 | 697 | -40% | 0 | 0 | — |
case-09 | pass→pass | 10,092 | 3,027 | -70% | 1 | 1 | 0% | 1,799 | 830 | -54% | 0 | 0 | — |
case-10 | pass→pass | 7,919 | 2,494 | -69% | 1 | 1 | 0% | 1,453 | 785 | -46% | 0 | 0 | — |
case-11 | fail→pass | 9,157 | 1,869 | -80% | 1 | 1 | 0% | 1,767 | 662 | -63% | 0 | 0 | — |
case-12 | fail→pass | 8,949 | 6,539 | -27% | 1 | 1 | 0% | 1,777 | 1,605 | -10% | 0 | 0 | — |
case-13 | fail→pass | 7,974 | 1,835 | -77% | 1 | 1 | 0% | 1,437 | 636 | -56% | 0 | 0 | — |
case-14 | fail→pass | 8,868 | 2,831 | -68% | 1 | 1 | 0% | 1,541 | 868 | -44% | 0 | 0 | — |
case-23 | fail→fail | 10,578 | 8,606 | -19% | 1 | 1 | 0% | 2,079 | 1,957 | -6% | 0 | 0 | — |
case-15 | fail→pass | 8,573 | 2,211 | -74% | 1 | 1 | 0% | 1,617 | 692 | -57% | 0 | 0 | — |
case-16 | pass→pass | 9,720 | 2,817 | -71% | 1 | 1 | 0% | 1,802 | 851 | -53% | 0 | 0 | — |
case-17 | pass→pass | 6,003 | 1,636 | -73% | 1 | 1 | 0% | 1,043 | 614 | -41% | 0 | 0 | — |
case-18 | fail→pass | 5,813 | 2,119 | -64% | 1 | 1 | 0% | 978 | 578 | -41% | 0 | 0 | — |
case-19 | fail→pass | 8,447 | 3,081 | -64% | 1 | 1 | 0% | 1,551 | 775 | -50% | 0 | 0 | — |
case-20 | pass→pass | 7,809 | 6,243 | -20% | 1 | 1 | 0% | 1,288 | 1,424 | +11% | 0 | 0 | — |
case-21 | fail→fail | 10,436 | 11,924 | +14% | 1 | 1 | 0% | 2,150 | 2,674 | +24% | 0 | 0 | — |
case-22 | fail→fail | 9,023 | 10,352 | +15% | 1 | 1 | 0% | 1,799 | 2,186 | +22% | 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. 23 cases were attempted. The headline lift of +43 percentage points is the difference between those two pass rates over the 23 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.