Juicebox SDK Patterns

Overview

Production-ready patterns for the Juicebox AI-powered people search API. Juicebox provides REST endpoints for searching professional profiles and enriching candidate data. The API authenticates via JUICEBOX_API_KEY and returns structured profile objects with LinkedIn URLs as natural dedup keys. A singleton client centralizes rate-limit handling across search and enrich endpoints.

Singleton Client

const JUICEBOX_BASE = 'https://api.juicebox.work/v1';
let _client: JuiceboxClient | null = null;
export function getClient(): JuiceboxClient {
  if (!_client) {
    const apiKey = process.env.JUICEBOX_API_KEY;
    if (!apiKey) throw new Error('JUICEBOX_API_KEY must be set — get it from juicebox.work/settings');
    _client = new JuiceboxClient(apiKey);
  }
  return _client;
}
class JuiceboxClient {
  private headers: Record<string, string>;
  constructor(apiKey: string) { this.headers = { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' }; }
  async search(query: string, limit = 20): Promise<SearchResponse> {
    const res = await fetch(`${JUICEBOX_BASE}/search`, {
      method: 'POST', headers: this.headers, body: JSON.stringify({ query, limit }) });
    if (!res.ok) throw new JuiceboxError(res.status, await res.text()); return res.json();
  }
  async enrich(linkedinUrl: string): Promise<Profile> {
    const res = await fetch(`${JUICEBOX_BASE}/enrich`, {
      method: 'POST', headers: this.headers, body: JSON.stringify({ linkedin_url: linkedinUrl }) });
    if (!res.ok) throw new JuiceboxError(res.status, await res.text()); return res.json();
  }
}

Error Wrapper

export class JuiceboxError extends Error {
  constructor(public status: number, message: string) { super(message); this.name = 'JuiceboxError'; }
}
export async function safeCall<T>(operation: string, fn: () => Promise<T>): Promise<T> {
  try { return await fn(); }
  catch (err: any) {
    if (err instanceof JuiceboxError && err.status === 429) { await new Promise(r => setTimeout(r, 5000)); return fn(); }
    if (err instanceof JuiceboxError && err.status === 401) throw new JuiceboxError(401, 'Invalid JUICEBOX_API_KEY');
    throw new JuiceboxError(err.status ?? 0, `${operation} failed: ${err.message}`);
  }
}

Request Builder

class JuiceboxSearchBuilder {
  private body: Record<string, any> = {};
  query(q: string) { this.body.query = q; return this; }
  limit(n: number) { this.body.limit = Math.min(n, 100); return this; }
  location(loc: string) { this.body.location = loc; return this; }
  title(t: string) { this.body.title_filter = t; return this; }
  company(c: string) { this.body.company_filter = c; return this; }
  yearsExp(min: number, max: number) { this.body.years_experience = { min, max }; return this; }
  build() { return this.body; }
}
// Usage: new JuiceboxSearchBuilder().query('ML engineer').location('San Francisco').yearsExp(3, 8).build();

Response Types

interface Profile {
  id: string; name: string; title: string; company: string;
  linkedin_url: string; location: string; skills: string[]; experience_years: number;
}
interface SearchResponse {
  profiles: Profile[]; total: number; has_more: boolean; cursor?: string;
}
interface EnrichResult {
  profile: Profile; education: Array<{ school: string; degree: string; year: number }>;
  experience: Array<{ company: string; title: string; start: string; end: string | null }>;
}

Testing Utilities

export function mockProfile(overrides: Partial<Profile> = {}): Profile {
  return { id: 'prof-001', name: 'Jane Smith', title: 'Senior ML Engineer',
    company: 'Acme Corp', linkedin_url: 'https://linkedin.com/in/janesmith',
    location: 'San Francisco, CA', skills: ['Python', 'PyTorch', 'MLOps'],
    experience_years: 6, ...overrides };
}
export function mockSearchResponse(count = 3): SearchResponse {
  return { profiles: Array.from({ length: count }, (_, i) => mockProfile({ id: `prof-${i}` })),
    total: count, has_more: false };
}

Error Handling

Pattern	When to Use	Example
safeCall wrapper	All Juicebox API calls	Structured error with operation context
Retry on 429	Batch search pipelines	5s backoff before retry
LinkedIn dedup	Multi-query search	Set<string> on linkedin_url prevents duplicates
Cursor pagination	Search results > 100	Pass cursor from previous response

Prerequisites

• A typed client boundary, secret-manager reference, sandbox workspace, approved source/destination allowlist, and synthetic prospect fixture.

Instructions

1. Construct clients from scoped environment configuration and reject unknown workspace, source, or destination before a request is sent.
2. Centralize source-authority, suppression, schema, redaction, and idempotency checks in the client boundary.
3. Model partial/denied responses explicitly; never turn a denied or unverified enrichment into a success or export fallback.
4. Test malformed data, rate limits, source/suppression rejection, and rollback using synthetic fixtures only.

Output

Produce a client-contract receipt with SDK revision, environment, source/destination policy revisions, fixture result, error classification, aggregate counts, export-count assertion, and rollback behavior. Exclude contacts, enrichment data, and credentials.

Examples

sdk=v3; env=sandbox; source=synthetic; destination=approved; suppression=pass; malformed=blocked; contacts_exported=0; telemetry=aggregate-only is a safe SDK result.

Resources

• Juicebox API Docs

Next Steps

Apply patterns in juicebox-core-workflow-a.