Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Install and configure Juicebox PeopleGPT API authentication. Use when setting up people search or initializing Juicebox. Trigger: "install juicebox", "setup juicebox", "juicebox auth".
.claude/skills/jeremylongshore-juicebox-install-auth/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -21% | 0% |
Set up Juicebox PeopleGPT API for AI-powered people search across 800M+ professional profiles.
bashnpm install @juicebox/sdk # or: pip install juicebox-sdk
bashexport JUICEBOX_API_KEY="jb_live_..." echo 'JUICEBOX_API_KEY=jb_live_your-key' >> .env
typescriptimport { JuiceboxClient } from '@juicebox/sdk'; const client = new JuiceboxClient({ apiKey: process.env.JUICEBOX_API_KEY }); const results = await client.search({ query: 'engineer', limit: 1 }); console.log(`Connected! ${results.total} profiles available`);
pythonfrom juicebox import JuiceboxClient client = JuiceboxClient(api_key=os.environ['JUICEBOX_API_KEY']) results = client.search(query='engineer', limit=1) print(f'Connected! {results.total} profiles')
| Error | Code | Solution | |-------|------|----------| | Invalid API key | 401 | Verify at app.juicebox.ai/settings | | Plan limit exceeded | 403 | Upgrade plan or check quota | | Rate limited | 429 | Check Retry-After header |
After auth, proceed to juicebox-hello-world.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 16,549 | 12,740 | -23% | 1 | 1 | 0% | 2,287 | 2,140 | -6% | 0 | 0 | — |
case-02 | fail→pass | 17,376 | 12,016 | -31% | 1 | 1 | 0% | 3,242 | 1,886 | -42% | 0 | 0 | — |
case-03 | fail→pass | 21,645 | 12,213 | -44% | 1 | 1 | 0% | 2,973 | 1,906 | -36% | 0 | 0 | — |
case-04 | fail→pass | 9,580 | 7,745 | -19% | 1 | 1 | 0% | 1,664 | 858 | -48% | 0 | 0 | — |
case-05 | fail→pass | 20,753 | 7,045 | -66% | 1 | 1 | 0% | 1,971 | 787 | -60% | 0 | 0 | — |
case-06 | pass→pass | 13,360 | 8,458 | -37% | 1 | 1 | 0% | 1,369 | 860 | -37% | 0 | 0 | — |
case-07 | pass→pass | 10,919 | 8,616 | -21% | 1 | 1 | 0% | 1,163 | 857 | -26% | 0 | 0 | — |
case-08 | fail→pass | 16,766 | 13,659 | -19% | 1 | 1 | 0% | 2,085 | 1,650 | -21% | 0 | 0 | — |
case-09 | pass→pass | 16,172 | 7,382 | -54% | 1 | 1 | 0% | 1,922 | 804 | -58% | 0 | 0 | — |
case-10 | pass→pass | 12,554 | 6,947 | -45% | 1 | 1 | 0% | 1,362 | 727 | -47% | 0 | 0 | — |
case-11 | fail→pass | 7,318 | 14,906 | +104% | 1 | 1 | 0% | 1,371 | 1,188 | -13% | 0 | 0 | — |
case-12 | fail→pass | 21,102 | 8,909 | -58% | 1 | 1 | 0% | 2,448 | 1,148 | -53% | 0 | 0 | — |
case-13 | fail→pass | 8,494 | 8,192 | -4% | 1 | 1 | 0% | 1,400 | 917 | -35% | 0 | 0 | — |
case-14 | fail→pass | 14,210 | 2,466 | -83% | 1 | 1 | 0% | 1,374 | 804 | -41% | 0 | 0 | — |
case-15 | fail→pass | 9,070 | 6,506 | -28% | 1 | 1 | 0% | 1,470 | 661 | -55% | 0 | 0 | — |
case-16 | fail→pass | 9,917 | 1,863 | -81% | 1 | 1 | 0% | 1,771 | 718 | -59% | 0 | 0 | — |
case-17 | fail→pass | 9,890 | 7,192 | -27% | 1 | 1 | 0% | 812 | 746 | -8% | 0 | 0 | — |
case-18 | fail→pass | 7,212 | 3,818 | -47% | 1 | 1 | 0% | 1,192 | 1,085 | -9% | 0 | 0 | — |
case-19 | fail→pass | 15,514 | 3,689 | -76% | 1 | 1 | 0% | 1,659 | 1,082 | -35% | 0 | 0 | — |
case-20 | pass→pass | 9,367 | 13,708 | +46% | 1 | 1 | 0% | 1,559 | 2,138 | +37% | 0 | 0 | — |
case-21 | pass→pass | 11,052 | 9,187 | -17% | 1 | 1 | 0% | 1,851 | 1,992 | +8% | 0 | 0 | — |
case-22 | pass→pass | 16,353 | 24,421 | +49% | 1 | 1 | 0% | 3,022 | 3,590 | +19% | 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 +64 percentage points is the difference between those two pass rates over the 22 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.