howoldisthisjob

howoldisthisjob estimates how old a job posting really is by collecting multiple signals, classifying them, and choosing the oldest credible posted date.

Current backend layers:

• JSON-LD JobPosting
• page metadata and regex extraction
• Open Graph / article metadata
• embedded JSON / hydration payloads
• Direct ATS fallbacks for Lever, Greenhouse, Ashby, SmartRecruiters, Workable, BambooHR, Brassring, SAP SuccessFactors, Rippling, iCIMS, Dover, Workday (CXS), Oracle HCM (recruitingCEJobRequisitionDetails), Jobvite (CompanyJobs XML), Avature, Gem, Teamtailor, Recruitee, Personio, Breezy HR, and JazzHR / applytojob
• Employer-specific URL resolvers that map into underlying platforms or a generic custom_backend bucket
• Platform detection with generic/archival fallbacks for ADP Workforce Now and Paycor/Newton
• Unsupported / early-stop detection for ClearCompany / HRMDirect, Google Careers, Indeed, and LinkedIn
• Jina render fallback for JS-heavy pages
• sitemap lastmod
• Wayback first-seen archive ceiling

Blocked / limited handling:

• Indeed -> blocked
• LinkedIn -> blocked
• Google Careers -> detected, but no reliable date exposed

CLI

python3 howoldisthisjob.py https://jobs.lever.co/skio/bbdd5a7b-652a-43ad-b92e-58f4e970c694

To run through npm:

npx howoldisthisjob https://jobs.lever.co/skio/bbdd5a7b-652a-43ad-b92e-58f4e970c694

To install the console scripts in a repo-local virtualenv:

python3 -m venv .venv
./.venv/bin/python -m pip install -e .
./.venv/bin/howoldisthisjob https://jobs.lever.co/skio/bbdd5a7b-652a-43ad-b92e-58f4e970c694

HTTP API

Run the local API server:

python3 howoldisthisjob_api.py --host 127.0.0.1 --port 8000

Or via the console script after editable install:

./.venv/bin/howoldisthisjob-api --host 127.0.0.1 --port 8000

Or through npm:

npx howoldisthisjob-api --host 127.0.0.1 --port 8000

Endpoints:

• GET /healthz
• GET /api/v1/estimate?url=<job-url>
• POST /api/v1/estimate with JSON body {"url": "<job-url>"}
• GET /api/v1/platforms — backend capability matrix (display name, supported flag, integration kind, detection patterns, notes) plus a summary count of direct vs generic vs blocked platforms

Response shape:

{
  "url": "https://jobs.lever.co/acme/123",
  "normalized_url": "https://jobs.lever.co/acme/123",
  "platform": "lever",
  "status": "success",
  "title": "Software Engineer",
  "company": "Acme",
  "location": "Remote",
  "employment_type": "Full-time",
  "likely_posted_date": "2024-01-01",
  "likely_age_days": 10,
  "confidence": "high",
  "reposted_likely": false,
  "summary": "Oldest credible posted date is 2024-01-01 from jsonld.jobposting.datePosted.",
  "chosen_source": {
    "date": "2024-01-01",
    "source": "jsonld.jobposting",
    "field": "datePosted",
    "kind": "posted",
    "reliability": "high"
  },
  "all_dates": [],
  "hidden_insights": {},
  "warnings": []
}

Cloudflare Deployment

This repo is set up to run the production API on Cloudflare:

• a Cloudflare Worker handles routing, CORS, and history endpoints
• a Cloudflare-managed container runs the existing Python API server
• D1 stores anonymous per-session search history
• the Worker is attached to api.howoldisthisjob.com

Expected production topology:

• Website: https://howoldisthisjob.com
• API: https://api.howoldisthisjob.com

CORS / cookie configuration:

• HOWOLDISTHISJOB_ALLOWED_ORIGINS controls allowed web origins for the Worker history routes.
• HOWOLDISTHISJOB_ALLOWED_EXTENSION_ORIGINS controls allowed Chrome extension origins for API and history routes. Keep both unpacked development IDs allowed while the old jobcarbon path is being phased out.
• the Python container keeps the same default allowlist used by the old Railway deployment path
• HOWOLDISTHISJOB_COOKIE_DOMAIN is optional; leaving it empty keeps the history cookie host-only on api.howoldisthisjob.com

Cloudflare files:

• Worker config: wrangler.jsonc
• Worker source: cloudflare/src/index.ts
• D1 schema: cloudflare/migrations/0001_history.sql
• Python container image: Dockerfile.cloudflare

Suggested Cloudflare flow:

1. npm install
2. npm run cf:types
3. wrangler d1 execute howoldisthisjob-history --remote --file=cloudflare/migrations/0001_history.sql
4. npm run cf:deploy

Testing

python3 -m unittest discover -s tests -v

The current suite covers:

• platform detection
• JSON-LD extraction
• retry behavior
• date normalization
• evidence ranking / repost detection
• rich result contract
• Greenhouse API fallback
• SmartRecruiters API fallback
• Rippling embedded __NEXT_DATA__ fallback
• iCIMS public /api/jobs fallback derived from the page's internal base host
• Dover application portal API fallback
• Workable embedded window.jobBoard payload fallback
• BambooHR /careers/{jobId}/detail JSON fallback
• Brassring DC.Date HTML metadata fallback
• SuccessFactors RSS + itemprop=datePosted fallback
• Workday CXS API fallback
• Oracle HCM recruitingCEJobRequisitionDetails fallback
• Jobvite CompanyJobs/Xml.aspx fallback
• Avature SearchJobs/feed/ + portal sitemap fallback
• Gem api.gem.com/job_board/v0/{board}/job_posts/ fallback
• Teamtailor public page metadata / JSON-LD fallback
• Recruitee /api/offers/{slug} fallback
• Personio page JSON-LD + /xml?language=... fallback
• Breezy HR data-position payload fallback
• JazzHR / applytojob JobPosting JSON-LD support
• Custom employer backend search.json?base_query={jobId} fallback
• Greenhouse resolver for employer-specific job URLs
• Oracle HCM resolver for employer-specific role URLs
• Custom employer backend ObjectID timestamp fallback
• /api/v1/platforms capability endpoint
• Jina render fallback
• sitemap and Wayback comparison evidence
• blocked-platform behavior