Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Always use 'npx convex dev' for development, never 'npx convex deploy' which is for production only
.claude/skills/kunanonj-cursor-plugin-convex-rule-use-convex-dev/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 9 |
| gemini-3.1-pro-preview | 100% | 1 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-21 | ✓→✓ | = Same ✓ | 3% | 0% |
| case-01 | ✓→✓ | = Same ✓ | 11% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 21% | 0% |
Important: When working with Convex, always use npx convex dev during development. Never use npx convex deploy unless you're deploying to production.
Use this for all development work:
bash# Start development (use this!) npx convex dev
Only use this when deploying to production:
bash# Production deployment only! npx convex deploy
Using deploy during development can:
Using dev for development ensures:
Development:
bash# 1. Start dev server npx convex dev # 2. Make changes to code # 3. Test locally # 4. Repeat
Deployment:
bash# Only after thorough testing: npx convex deploy # Or in CI/CD: # - Run tests # - If tests pass, deploy
bashnpm install convex npx convex dev # Creates account & starts dev
bashnpx convex dev # Always use this # Keep it running while you code
bash# Test everything first! npm run test npm run build # Then deploy npx convex deploy
yaml# GitHub Actions example - name: Deploy to Convex run: npx convex deploy --cmd 'npm run build' env: CONVEX_DEPLOY_KEY: ${{ secrets.CONVEX_DEPLOY_KEY }}
If you see someone suggest:
npx convex deploy to test" ❌ Wrong!deploy for development" ❌ Wrong!Always use npx convex dev for development!
When writing docs or instructions:
npx convex dev for developmentnpx convex deploy as "Production only"| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | pass→pass | 5,494 | 2,705 | -51% | 1 | 1 | 0% | 1,078 | 1,108 | +3% | 0 | 0 | — |
case-01 | pass→pass | 4,784 | 2,539 | -47% | 1 | 1 | 0% | 1,033 | 1,142 | +11% | 0 | 0 | — |
case-02 | pass→pass | 9,068 | 8,012 | -12% | 1 | 1 | 0% | 1,789 | 2,173 | +21% | 0 | 0 | — |
case-03 | pass→pass | 2,956 | 2,691 | -9% | 1 | 1 | 0% | 599 | 1,144 | +91% | 0 | 0 | — |
case-04 | pass→pass | 9,647 | 4,514 | -53% | 1 | 1 | 0% | 1,867 | 1,632 | -13% | 0 | 0 | — |
case-05 | pass→pass | 9,518 | 6,120 | -36% | 1 | 1 | 0% | 1,564 | 1,825 | +17% | 0 | 0 | — |
case-06 | pass→pass | 10,372 | 7,150 | -31% | 1 | 1 | 0% | 2,162 | 2,050 | -5% | 0 | 0 | — |
case-07 | pass→pass | 8,581 | 6,723 | -22% | 1 | 1 | 0% | 1,609 | 1,823 | +13% | 0 | 0 | — |
case-08 | fail→pass | 12,183 | 6,876 | -44% | 1 | 1 | 0% | 2,258 | 2,055 | -9% | 0 | 0 | — |
case-09 | pass→pass | 10,549 | 4,530 | -57% | 1 | 1 | 0% | 2,031 | 1,496 | -26% | 0 | 0 | — |
case-10 | pass→pass | 2,232 | 1,423 | -36% | 1 | 1 | 0% | 433 | 939 | +117% | 0 | 0 | — |
case-11 | pass→pass | 7,496 | 4,352 | -42% | 1 | 1 | 0% | 1,537 | 1,503 | -2% | 0 | 0 | — |
case-12 | pass→pass | 12,121 | 9,591 | -21% | 1 | 1 | 0% | 2,449 | 2,535 | +4% | 0 | 0 | — |
case-13 | pass→pass | 8,804 | 8,381 | -5% | 1 | 1 | 0% | 1,793 | 2,177 | +21% | 0 | 0 | — |
case-14 | pass→pass | 9,325 | 7,346 | -21% | 1 | 1 | 0% | 1,677 | 1,980 | +18% | 0 | 0 | — |
case-15 | fail→pass | 7,159 | 4,035 | -44% | 1 | 1 | 0% | 1,527 | 1,402 | -8% | 0 | 0 | — |
case-16 | pass→pass | 6,902 | 3,755 | -46% | 1 | 1 | 0% | 1,387 | 1,362 | -2% | 0 | 0 | — |
case-17 | pass→pass | 10,543 | 5,953 | -44% | 1 | 1 | 0% | 2,210 | 1,845 | -17% | 0 | 0 | — |
case-18 | pass→pass | 12,311 | 11,804 | -4% | 1 | 1 | 0% | 2,191 | 2,689 | +23% | 0 | 0 | — |
case-19 | pass→pass | 6,961 | 5,222 | -25% | 1 | 1 | 0% | 1,312 | 1,514 | +15% | 0 | 0 | — |
case-20 | pass→pass | 2,478 | 2,957 | +19% | 1 | 1 | 0% | 410 | 1,173 | +186% | 0 | 0 | — |
case-22 | pass→pass | 4,254 | 3,856 | -9% | 1 | 1 | 0% | 935 | 1,326 | +42% | 0 | 0 | — |
case-23 | pass→pass | 10,306 | 5,590 | -46% | 1 | 1 | 0% | 2,140 | 1,767 | -17% | 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 +9 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.