Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Execute use when deploying Genkit applications to production with Terraform. Trigger with phrases like "deploy genkit terraform", "provision genkit infrastructure", "firebase functions terraform", "cloud run deployment", or "genkit production infrastructure". Provisions Firebase Functions, Cloud Run services, GKE clusters, monitoring dashboards, and CI/CD for AI workflows.
.claude/skills/jeremylongshore-genkit-infra-expert/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -27% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 11% | 0% |
| case-15 | ✓→✓ | ＝ Same ✓ | 21% | 0% |
Deploy Genkit applications to production with Terraform (Firebase Functions, Cloud Run, or GKE) with secure secrets handling and observability. Use this skill to choose a target, generate the Terraform baseline, wire up Secret Manager, and provide a validation checklist for your Genkit flows.
Before using this skill, ensure:
See Terraform implementation details for output format specifications.
See ${CLAUDE_SKILL_DIR}/references/errors.md for comprehensive error handling.
See ${CLAUDE_SKILL_DIR}/references/examples.md for detailed examples.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | 23,119 | 24,598 | +6% | 1 | 1 | 0% | 4,952 | 6,353 | +28% | 0 | 0 | — |
case-01 | fail→fail | 35,091 | 28,397 | -19% | 1 | 1 | 0% | 4,956 | 5,473 | +10% | 0 | 0 | — |
case-02 | fail→fail | 45,012 | 48,152 | +7% | 1 | 1 | 0% | 6,229 | 7,615 | +22% | 0 | 0 | — |
case-03 | fail→fail | 43,406 | 33,770 | -22% | 1 | 1 | 0% | 8,239 | 6,860 | -17% | 0 | 0 | — |
case-04 | fail→fail | 33,276 | 29,526 | -11% | 1 | 1 | 0% | 4,277 | 5,499 | +29% | 0 | 0 | — |
case-05 | pass→pass | 47,828 | 30,474 | -36% | 1 | 1 | 0% | 7,328 | 5,318 | -27% | 0 | 0 | — |
case-06 | fail→fail | 30,654 | 27,518 | -10% | 1 | 1 | 0% | 4,376 | 5,477 | +25% | 0 | 0 | — |
case-07 | pass→pass | 23,945 | 18,162 | -24% | 1 | 1 | 0% | 3,808 | 4,219 | +11% | 0 | 0 | — |
case-08 | fail→fail | 46,426 | 30,598 | -34% | 1 | 1 | 0% | 6,509 | 5,568 | -14% | 0 | 0 | — |
case-10 | fail→fail | 22,257 | 15,864 | -29% | 1 | 1 | 0% | 3,452 | 3,546 | +3% | 0 | 0 | — |
case-11 | fail→fail | 34,365 | 30,960 | -10% | 1 | 1 | 0% | 5,395 | 6,141 | +14% | 0 | 0 | — |
case-12 | fail→fail | 31,484 | 25,423 | -19% | 1 | 1 | 0% | 5,504 | 6,198 | +13% | 0 | 0 | — |
case-13 | fail→pass | 33,333 | 28,194 | -15% | 1 | 1 | 0% | 5,794 | 4,983 | -14% | 0 | 0 | — |
case-14 | fail→fail | 27,057 | 31,101 | +15% | 1 | 1 | 0% | 4,528 | 6,456 | +43% | 0 | 0 | — |
case-15 | pass→pass | 28,873 | 29,203 | +1% | 1 | 1 | 0% | 4,630 | 5,617 | +21% | 0 | 0 | — |
case-16 | fail→fail | 29,189 | 40,116 | +37% | 1 | 1 | 0% | 3,896 | 3,741 | -4% | 0 | 0 | — |
case-17 | fail→fail | 23,119 | 25,233 | +9% | 1 | 1 | 0% | 3,504 | 4,560 | +30% | 0 | 0 | — |
case-18 | fail→fail | 32,598 | 36,728 | +13% | 1 | 1 | 0% | 5,030 | 7,461 | +48% | 0 | 0 | — |
case-19 | fail→fail | 20,734 | 20,956 | +1% | 1 | 1 | 0% | 3,241 | 4,982 | +54% | 0 | 0 | — |
case-20 | pass→pass | 13,567 | 11,777 | -13% | 1 | 1 | 0% | 1,783 | 1,842 | +3% | 0 | 0 | — |
case-21 | pass→pass | 12,639 | 7,555 | -40% | 1 | 1 | 0% | 1,452 | 1,784 | +23% | 0 | 0 | — |
case-22 | pass→pass | 17,002 | 15,416 | -9% | 1 | 1 | 0% | 2,536 | 3,783 | +49% | 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 +9 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.