Talent Sourcer (live)

Endpoints it uses

• /api/v1/companies/entity-id10 credits/call
• /api/v1/search/people-live10 credits/call
• /api/v1/profile/full10 credits/call

Ballpark: ~300 credits for 10 typical runs. Signup includes 300 free credits.

What's inside (full preview)

---
name: linkedin-talent-sourcer
description: Use when the user wants to find people by company history or location — current employees of a company, ALUMNI who used to work somewhere, people from a specific school, or professionals in a specific city. Triggers on "who works at X", "find people who left X", "ex-Google engineers in Tel Aviv", "alumni of this company", "sourcing list by employer". This is the only path that filters by PAST employer.
---

# Talent Sourcer (via Zooq live search)

You source people through Zooq's LIVE people search — the one surface that can filter by current company, **past company**, and school together. "Ex-employees of X" is this skill's signature move; no other endpoint can express it.

## Inputs you need from the user

- **The filter that matters**: a company (for current or past employees), a school, a city/country, a title keyword — any combination. Ask for whichever single one they lead with; don't interrogate.
- **Count** (optional, default 20).

## Step 1 — resolve company slugs to ids (only if filtering by company)

`currentCompany` / `pastCompany` take NUMERIC organization ids. Resolve each company once:

```
GET https://zooq.dev/api/v1/companies/entity-id?slug=<COMPANY_SLUG>
Headers:
  X-API-Key: REPLACE_WITH_YOUR_KEY
```

The slug is the part after `linkedin.com/company/`. If the user gave a company NAME, find the slug first via `GET /api/v1/companies/name-lookup?name=<NAME>` and read `data[].slug`. Resolve once, reuse the id.

## Step 2 — the geo id (only if filtering by place)

`geoEntityId` takes LinkedIn's own public geo id, and it works — verified. To find one: type the place into the location box on LinkedIn search, pick the auto-complete match, press enter, and read `geoId=` out of the browser address bar. Example: Tel Aviv-Yafo is `101570771`. Prefer a CITY id over a country id. If the user can't fetch one, skip the filter and post-filter on the results' location text instead.

## Step 3 — search

```
GET https://zooq.dev/api/v1/search/people-live?pastCompany=<ORG_ID>&geoEntityId=<GEO_ID>&count=20
Headers:
  X-API-Key: REPLACE_WITH_YOUR_KEY
```

Useful params (all optional, comma-separate multiples): `keyword`, `title`, `firstName`, `lastName`, `currentCompany`, `pastCompany`, `school`, `geoEntityId`, `profileLanguage`, `count` (max 50), `start` (pagination, 0–999). Do NOT pass `industry` — it takes an id nothing currently produces, and free text is ignored.

10 credits per page. `data.total` tells you the pool size; page with `start` only as far as the user actually needs.

## Step 4 — enrich the shortlist (optional, default top 5)

For the people worth a closer look, pull the full profile — 10 credits each, so shortlist first:

```
GET https://zooq.dev/api/v1/profile/full?handle=<HANDLE>
Headers:
  X-API-Key: REPLACE_WITH_YOUR_KEY
```

## Output format

A ranked table: Name · Headline · Location · Why they match (one concrete phrase — "3y at <company>, left 2025", not a score). Then one line: pool size, pages fetched, credits spent. Offer the natural next step — enrich more of the list, or export it (the Enrich & Save skill picks up exactly where this leaves off).

## Failure rules

- Zero results with filters ≠ zero results, period. Drop the narrowest filter (usually the title keyword) and retry once before reporting empty.
- Never present the deduplicated-dataset search (`/api/v1/search/people`) as interchangeable — it has no past-company filter. It IS the better tool when the user wants plain-text city/country filtering with firmographics; say so if their ask fits it better.
- Never fabricate profiles. Every row you show came from a real response.