Job Board Aggregator — Indeed, LinkedIn, Glassdoor & ZipRecruiter API for Recruiting & Lead-Gen Agents

Job-market data API that scrapes Indeed, Glassdoor, ZipRecruiter, and LinkedIn in one call and returns typed JSON rows — schema below. Every listing is normalized to a single 35-field contract (salary range + annualized estimate, skills, experience level, education requirement, benefits) and deduplicated across boards. Built for recruiting, lead-gen, and market-research agents that need multi-board coverage without stitching four scrapers together.

Why agents use this actor

• Deterministic typed output — every row follows the dataset schema below; stable id hash per posting, consistent field names regardless of source board.
• Multi-board aggregation in one call — one input, four job boards, cross-board deduplication. No fan-out orchestration on your side.
• Parsed intelligence, not raw HTML — salary min/max/period/currency plus estimatedAnnualSalary, technicalSkills/softSkills extraction, experienceLevel, yearsExperienceRequired, educationRequired.
• Clear error semantics — invalid input or zero results exits non-zero with a status message; per-site failures are reported in the SUMMARY key-value record, never as billable dataset rows.
• No auth or cookies needed — public listings only; per-site resilience built in (scrapeIndividually isolates board failures).

Input schema

Field	Type	Required	Default	Description
searchTerm	string	yes	— (prefill: software engineer)	Job title or keywords to search for
location	string	yes	— (prefill: Austin, TX)	City, state, or country for job search
sites	array	no	["glassdoor", "zip_recruiter", "linkedin"]	Boards to scrape. Allowed: indeed, glassdoor, zip_recruiter, linkedin
resultsPerSite	integer	no	25	Max results per site (1–100)
jobType	string	no	""	Filter: fulltime, parttime, contract, internship, or "" for all
isRemote	boolean	no	false	Only return remote positions
distanceMiles	integer	no	50	Search radius from location (5–100)
hoursOld	integer	no	168	Only jobs posted within this many hours (1–720)
countryIndeed	string	no	usa	Indeed country: usa, uk, canada, germany, france, india, australia
scrapeIndividually	boolean	no	true	Scrape each board separately so one board's failure doesn't kill the run
enableDeduplication	boolean	no	true	Remove duplicate listings across boards (title+company+location hash)
extractSkills	boolean	no	true	Extract technical and soft skills from descriptions
proxyConfiguration	object	no	Apify residential proxy	Proxy settings. Residential is the default because Glassdoor and LinkedIn block datacenter IPs; residential traffic is billed to your Apify account

Output schema

Single mode: every dataset row is one normalized job listing with all fields below. Fields the source board doesn't provide are null (or empty string / empty array), never omitted.

Field	Type	Nullable	Description
id	string	no	Stable 16-char hash of title+company+location
source	string	no	indeed, glassdoor, zip_recruiter, or linkedin
title	string	no	Job title
company	string	no	Company name
location	string	no	Location as posted
jobType	string	yes	fulltime, parttime, contract, internship, or null
isRemote	boolean	no	Board-flagged remote position
experienceLevel	string	yes	entry, mid, senior, executive, or null (inferred)
yearsExperienceRequired	integer	yes	Min years parsed from description
educationRequired	string	yes	HighSchool, Associates, Bachelors, Masters, PhD, or null
salaryMin	number	yes	Posted salary lower bound
salaryMax	number	yes	Posted salary upper bound
salaryPeriod	string	yes	hourly, weekly, monthly, yearly, or null
salaryCurrency	string	no	Currency code (default USD)
estimatedAnnualSalary	number	yes	Range midpoint annualized from salaryPeriod
technicalSkills	array<string>	no	Detected tech skills (empty if extractSkills false)
softSkills	array<string>	no	Detected soft skills
allSkills	array<string>	no	technicalSkills + softSkills
skillsCount	integer	no	Total skills detected
description	string	no	Description text, truncated to 5000 chars (often empty on LinkedIn)
descriptionLength	integer	no	Full description length in chars
requirements	array<string>	no	Parsed requirement phrases (max 10)
benefits	array<string>	no	Detected benefit keywords (401k, pto, equity, ...)
companyLogo	string	yes	Logo URL
companyRating	number	yes	Company rating (mainly Glassdoor)
companyIndustry	string	yes	Industry
companySize	string	yes	Headcount range
companyUrl	string	yes	Company profile URL on the source board
postedDate	string	no	YYYY-MM-DD (empty string when unknown)
scrapedAt	string	no	ISO 8601 UTC scrape timestamp
applyUrl	string	no	Direct URL to the posting
sourceUrl	string	no	Same as applyUrl
hasRemoteOption	boolean	no	Description mentions remote / work from home
hasVisaSponsorship	boolean	no	Description mentions visa / sponsorship
isEasyApply	boolean	no	Board-flagged one-click apply

Real example row (from a live run):

{
  "id": "98ffae5ccb0ec729",
  "source": "linkedin",
  "title": "Software Engineer, Realtime",
  "company": "Cloudflare",
  "location": "Austin, TX",
  "jobType": null,
  "isRemote": false,
  "experienceLevel": null,
  "yearsExperienceRequired": null,
  "educationRequired": null,
  "salaryMin": null,
  "salaryMax": null,
  "salaryPeriod": null,
  "salaryCurrency": "USD",
  "estimatedAnnualSalary": null,
  "technicalSkills": [],
  "softSkills": [],
  "allSkills": [],
  "skillsCount": 0,
  "description": "",
  "descriptionLength": 0,
  "requirements": [],
  "benefits": [],
  "companyLogo": null,
  "companyRating": null,
  "companyIndustry": null,
  "companySize": null,
  "companyUrl": "https://www.linkedin.com/company/cloudflare",
  "postedDate": "2026-08-13",
  "scrapedAt": "2026-08-15T01:34:22.114155",
  "applyUrl": "https://www.linkedin.com/jobs/view/4441070330",
  "sourceUrl": "https://www.linkedin.com/jobs/view/4441070330",
  "hasRemoteOption": false,
  "hasVisaSponsorship": false,
  "isEasyApply": false
}

Error semantics

• Invalid input (empty searchTerm/location, empty sites): the run exits with code 1 and a status message explaining what's missing. No dataset items.
• Zero results (all boards blocked, errored, or no matching jobs): the run exits with code 1 and a per-site status message — never a silent SUCCEEDED with 0 items. Retry with different boards, a broader search, or proxies.
• Partial failure (some boards succeed, some fail): the run SUCCEEDS with the rows it got. Per-site outcomes (success / empty / error + error text) are in the SUMMARY key-value record (also stored as summary for backward compatibility) under siteStatus. Failed sites never produce dataset rows, so you're never billed for errors.
• Retry guidance for agents: check SUMMARY.siteStatus; a 403 on Glassdoor means enable residential proxy; LinkedIn empty/401 is rate limiting — retry later or lower resultsPerSite.

Use from AI agents (MCP)

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com/?tools=apricot_blackberry/job-board-aggregator",
      "headers": { "Authorization": "Bearer <YOUR_APIFY_TOKEN>" }
    }
  }
}

Works in Claude, Cursor, ChatGPT deep research connectors, and any MCP client; the input schema above is the tool's parameter schema.

Use from code

curl:

curl -X POST "https://api.apify.com/v2/acts/apricot_blackberry~job-board-aggregator/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"searchTerm": "software engineer", "location": "Austin, TX", "sites": ["glassdoor", "zip_recruiter", "linkedin"], "resultsPerSite": 10}'

JavaScript (apify-client):

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('apricot_blackberry/job-board-aggregator').call({
    searchTerm: 'software engineer',
    location: 'Austin, TX',
    sites: ['glassdoor', 'zip_recruiter', 'linkedin'],
    resultsPerSite: 10,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();

Python (apify_client):

from apify_client import ApifyClient

client = ApifyClient(os.environ["APIFY_TOKEN"])
run = client.actor("apricot_blackberry/job-board-aggregator").call(run_input={
    "searchTerm": "software engineer",
    "location": "Austin, TX",
    "sites": ["glassdoor", "zip_recruiter", "linkedin"],
    "resultsPerSite": 10,
})
items = client.dataset(run["defaultDatasetId"]).list_items().items

Use from automation platforms

• n8n / Make / Zapier: use the native Apify integration and choose job-board-aggregator by name.
• LangChain / LlamaIndex: use the Apify tool wrappers (ApifyWrapper / ApifyActor) with the same input JSON.
• Webhooks: configure an Apify webhook to fire on run completion (ACTOR.RUN.SUCCEEDED/FAILED) to push new listings into your ATS or CRM pipeline without polling.

Supported job sites

• Indeed — largest job board, best coverage, most reliable (7+ countries via countryIndeed)
• Glassdoor — includes company ratings; aggressive anti-bot, needs residential proxy
• ZipRecruiter — good for SMB jobs
• LinkedIn — professional network jobs; strictest rate limits (~10 pages per IP), list endpoint returns no descriptions

Rate limits & best practices

• LinkedIn has the strictest rate limits; Indeed is the most reliable with minimal rate limiting.
• Recommended: 25–50 results per site per run; use scheduled runs for continuous monitoring.
• Keep scrapeIndividually on so one blocked board doesn't sink the run.

Pricing

$0.10 start + $0.05 per job listing delivered (PAY_PER_EVENT). A default run (3 sites × 25 results/site, deduplicated) typically lands around $2–4; residential proxy usage (the default) is billed to your Apify account on top of this.

FAQ

Why are some fields null/empty? Boards report different data. LinkedIn's list endpoint returns no descriptions, so description-derived fields (skills, requirements, education) are empty for LinkedIn rows.

How does deduplication work? MD5 hash of lowercased title+company+location(first 20 chars); first occurrence wins. Duplicate count is in the SUMMARY record.

Do I need an account on the job boards? No — public listings only, no login or cookies.

Changelog

v2.1 (2026-08)

• Output dataset schema published (.actor/dataset_schema.json) — typed, agent-readable contract for all 35 fields.
• Agent integration docs: MCP server config, curl/JS/Python snippets.
• indeed added to the sites input enum (was documented but not selectable).
• Fail-loud semantics: invalid input or zero total results now exits with code 1 and a descriptive status message instead of silently succeeding with 0 items.
• Run summary now also stored under the standard SUMMARY key-value record (existing summary record kept for backward compatibility).

v2.0

• Per-site individual scraping (scrapeIndividually) with per-site success/failure status.
• Proxy rotation support; skill/requirement/benefit extraction; cross-board deduplication.