Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Configure Juicebox team access. Trigger: "juicebox rbac", "juicebox team roles".
.claude/skills/jeremylongshore-juicebox-enterprise-rbac/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -46% | 0% |
Juicebox provides AI-powered people search and analysis for recruiting and sales teams. Enterprise RBAC controls who can search candidate databases, enrich profiles with contact info, trigger outreach sequences, and export data. Workspace admins manage team seats and API usage limits. Analysts run searches but may be restricted from exporting PII. Viewers can review saved searches without accessing raw contact data. SOC 2 compliance requires audit logging on all data enrichment and export actions.
| Role | Permissions | Scope | |------|------------|-------| | Workspace Admin | Manage seats, billing, API keys, configure integrations | Entire workspace | | Recruiter | Search, enrich, access contact info, run outreach campaigns | All datasets | | Analyst | Search and enrich profiles, view analytics dashboards | Assigned datasets | | Sourcer | Search and enrich only, no contact reveal or outreach | Assigned datasets | | Viewer | View saved searches and reports, no data access or export | Read-only |
typescriptasync function checkJuiceboxAccess(userId: string, action: string, datasetId: string): Promise<boolean> { const response = await fetch(`${JUICEBOX_API}/v1/workspaces/${WORKSPACE_ID}/permissions`, { headers: { Authorization: `Bearer ${JUICEBOX_API_KEY}`, 'Content-Type': 'application/json' }, }); const perms = await response.json(); const user = perms.members.find((m: any) => m.id === userId); if (!user) return false; const rolePerms = ROLE_PERMISSIONS[user.role]; return rolePerms?.[action] && (user.datasets.includes(datasetId) || user.role === 'admin'); }
typescriptasync function assignWorkspaceRole(email: string, role: string, datasets: string[]): Promise<void> { await fetch(`${JUICEBOX_API}/v1/workspaces/${WORKSPACE_ID}/members`, { method: 'POST', headers: { Authorization: `Bearer ${JUICEBOX_API_KEY}`, 'Content-Type': 'application/json' }, body: JSON.stringify({ email, role, datasetAccess: datasets }), }); } async function revokeAccess(email: string): Promise<void> { await fetch(`${JUICEBOX_API}/v1/workspaces/${WORKSPACE_ID}/members/${email}`, { method: 'DELETE', headers: { Authorization: `Bearer ${JUICEBOX_API_KEY}` }, }); }
typescriptinterface JuiceboxAuditEntry { timestamp: string; userId: string; role: string; action: 'search' | 'enrich' | 'contact_reveal' | 'export' | 'outreach' | 'role_change'; datasetId: string; recordCount?: number; result: 'allowed' | 'denied'; } function logAccess(entry: JuiceboxAuditEntry): void { console.log(JSON.stringify({ ...entry, workspaceId: process.env.JUICEBOX_WORKSPACE_ID })); }
| Issue | Cause | Fix | |-------|-------|-----| | 403 on contact reveal endpoint | User role lacks contact permission | Upgrade to recruiter role or request admin approval | | Export returns empty results | Dataset not assigned to user | Add dataset to user's access list in workspace settings | | API rate limit exceeded | Too many enrichment calls per hour | Check role-based rate limits, batch requests | | Saved search not visible | Search created in different dataset scope | Verify dataset access or share search explicitly | | Seat limit reached | All workspace licenses used | Remove inactive members or upgrade plan |
See juicebox-security-basics.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | pass→pass | 20,959 | 19,095 | -9% | 1 | 1 | 0% | 2,637 | 3,399 | +29% | 0 | 0 | — |
case-01 | fail→pass | 22,599 | 20,233 | -10% | 1 | 1 | 0% | 3,630 | 4,483 | +23% | 0 | 0 | — |
case-02 | fail→pass | 16,939 | 15,967 | -6% | 1 | 1 | 0% | 2,433 | 3,237 | +33% | 0 | 0 | — |
case-03 | fail→pass | 19,673 | 14,557 | -26% | 1 | 1 | 0% | 2,414 | 2,639 | +9% | 0 | 0 | — |
case-04 | pass→fail | 20,502 | 21,405 | +4% | 1 | 1 | 0% | 2,822 | 4,048 | +43% | 0 | 0 | — |
case-05 | pass→pass | 17,252 | 16,221 | -6% | 1 | 1 | 0% | 1,977 | 2,923 | +48% | 0 | 0 | — |
case-07 | pass→pass | 16,124 | 8,233 | -49% | 1 | 1 | 0% | 1,791 | 1,514 | -15% | 0 | 0 | — |
case-08 | pass→pass | 16,195 | 12,119 | -25% | 1 | 1 | 0% | 1,684 | 1,884 | +12% | 0 | 0 | — |
case-09 | pass→fail | 16,818 | 7,295 | -57% | 1 | 1 | 0% | 1,814 | 1,461 | -19% | 0 | 0 | — |
case-10 | fail→pass | 17,365 | 14,041 | -19% | 1 | 1 | 0% | 2,391 | 2,902 | +21% | 0 | 0 | — |
case-11 | fail→pass | 27,614 | 8,057 | -71% | 1 | 1 | 0% | 2,925 | 1,569 | -46% | 0 | 0 | — |
case-12 | fail→pass | 17,088 | 8,565 | -50% | 1 | 1 | 0% | 1,641 | 1,375 | -16% | 0 | 0 | — |
case-13 | fail→pass | 13,569 | 8,599 | -37% | 1 | 1 | 0% | 1,335 | 1,447 | +8% | 0 | 0 | — |
case-14 | pass→pass | 18,809 | 3,866 | -79% | 1 | 1 | 0% | 1,861 | 1,615 | -13% | 0 | 0 | — |
case-15 | fail→pass | 21,565 | 9,796 | -55% | 1 | 1 | 0% | 2,525 | 2,969 | +18% | 0 | 0 | — |
case-20 | pass→pass | 15,324 | 13,420 | -12% | 1 | 1 | 0% | 2,244 | 2,184 | -3% | 0 | 0 | — |
case-16 | fail→pass | 11,965 | 4,218 | -65% | 1 | 1 | 0% | 1,902 | 1,731 | -9% | 0 | 0 | — |
case-17 | fail→pass | 9,676 | 9,219 | -5% | 1 | 1 | 0% | 1,893 | 1,868 | -1% | 0 | 0 | — |
case-18 | fail→pass | 20,220 | 9,159 | -55% | 1 | 1 | 0% | 2,056 | 1,554 | -24% | 0 | 0 | — |
case-19 | pass→pass | 19,773 | 2,851 | -86% | 1 | 1 | 0% | 2,202 | 1,411 | -36% | 0 | 0 | — |
case-21 | fail→pass | 17,831 | 5,135 | -71% | 1 | 1 | 0% | 1,615 | 1,920 | +19% | 0 | 0 | — |
case-22 | fail→fail | 18,727 | 2,417 | -87% | 1 | 1 | 0% | 2,576 | 1,428 | -45% | 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 +45 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.