Job Tracker

Scrapes Senior PM-and-above product roles (Israel + Remote) daily from a list of target companies, tracks open/closed status, displays results in a Claude artifact dashboard.

What's tracked

Automated (works end-to-end via GitHub Actions cron):

Company	ATS	Reliability
Melio	Greenhouse API	✅ High
Forter	Greenhouse API	✅ High
Riskified	Greenhouse API	✅ High
HoneyBook	Greenhouse API	✅ High
Amazon	Amazon Jobs API	✅ High
Monday.com	Custom HTML	⚠️ Best-effort
AppsFlyer	Custom HTML	⚠️ Best-effort
Fiverr	Custom HTML	⚠️ Best-effort
Lightricks	Custom HTML	⚠️ Best-effort
Papaya Global	Custom HTML	⚠️ Best-effort
JFrog	Custom HTML	⚠️ Best-effort

"Best-effort" means the parser uses the Comeet job UID regex (XX.XXX format) to find jobs in HTML. Works as long as the careers page exposes those UIDs in <a href> attributes. If a company redesigns their careers site or hides links behind JavaScript, the scraper silently returns 0 jobs and logs an error rather than mass-closing existing entries (safety guard).

Manual checking required (weekly):

Company	Why skipped
Wix, Google, Microsoft, Meta, Salesforce, Intuit, Apple, Nvidia	JS-rendered or Workday-based SPAs; need browser automation
Lemonade	Custom careers site, not Greenhouse-public
Snyk	JS-rendered SPA at snyk.io/careers
Rapyd, Gong, Tipalti	Likely use Greenhouse but their token isn't obvious; verify by trying boards.greenhouse.io/<guess>

For these, set up LinkedIn job alerts as a fallback — LinkedIn scrapes most of them and emails you on new postings.

Architecture

GitHub repo                                  Claude artifact
┌─────────────────────────────────────┐      ┌────────────────────┐
│ .github/workflows/scrape-daily.yml  │      │ job_tracker_       │
│   (cron, 07:00 UTC daily)           │      │ dashboard.jsx      │
│           │                          │      │                    │
│           ▼                          │      │  Reads jobs.json   │
│ scraper.py  ──► Greenhouse API      │      │  on Refresh        │
│              ──► Comeet HTML parse  │ ───► │                    │
│              ──► Amazon Jobs API    │ fetch│  Stores locally    │
│           │                          │ raw  │  in window.storage │
│           ▼                          │      │                    │
│ data/jobs.json  (committed back)     │      │  Cards, filters,   │
└─────────────────────────────────────┘      │  detail modal      │
                                              └────────────────────┘

Setup

1. Create a GitHub repo

Public repo (so raw URL works without auth). Drop in:

job-tracker/
├── README.md
├── companies.json
├── scraper.py
├── requirements.txt
├── data/
│   └── jobs.json          ← starts as {"lastUpdated": null, "jobs": []}
└── .github/
    └── workflows/
        └── scrape-daily.yml

2. Enable workflow write permissions

Repo → Settings → Actions → General → Workflow permissions → Read and write permissions → Save.

3. First run

Actions tab → "Scrape jobs daily" → "Run workflow" → manual trigger. Wait ~1 min. Check data/jobs.json got updated.

4. Wire up dashboard

In data/jobs.json, click "Raw" button, copy URL:

https://raw.githubusercontent.com/<you>/job-tracker/main/data/jobs.json

Open the dashboard artifact, gear icon, paste URL, Save, Refresh.

Updating filters

In companies.json under filters:

• titleMustMatch — must contain "product"
• seniorityIndicators — at least one of: senior, sr, staff, principal, lead, group, director, vp, head, chief
• excludeTitle — excludes: marketing, designer, engineer, analyst, owner, ops, etc.
• locationMustMatch — Israeli cities OR "remote"
• excludeLocation — excludes "Remote - US", "Remote India", etc.

Edit and commit. Next run re-applies filters.

Adding more companies

If they use Greenhouse: Visit boards.greenhouse.io/<your-guess>. If the page loads, add:

{ "name": "CompanyName", "ats": "greenhouse", "token": "your-guess" }

If they use Comeet (job URLs contain XX.XXX patterns): Add:

{ "name": "CompanyName", "ats": "comeet_html", "url": "https://company.com/careers" }

Custom HTML site: You'll need to add a new fetcher function in scraper.py and register it in the FETCHERS dict at the top.

Verifying unconfirmed Greenhouse tokens

Three companies (Rapyd, Gong, Tipalti) probably use Greenhouse but their slug isn't obvious. Quick way to verify:

1. Open https://boards.greenhouse.io/<guess> in browser
2. If the careers page loads → that slug works → update companies.json with "ats": "greenhouse", "token": "<guess>"
3. If 404 → try variations (e.g., gonghire, rapydpayments, tipaltihq)

Once a token works, the next workflow run will pick up that company's jobs automatically.

Limitations

• Comeet HTML parsing is fragile: relies on the careers page exposing job links with the Comeet UID pattern. If a company changes their careers page design, the parser may silently return 0 jobs — logged in fetchErrors, not catastrophic.
• Title/location parsing for Comeet jobs: works well when title/location are in separate HTML elements (the common case). When everything's concatenated in one text node, the parser does its best using a regex for trailing "City, CC" patterns; titles with embedded acronyms (like "VP Product") can get split incorrectly.
• No detailed description for Comeet jobs: the scraper only grabs title + location from the listing page, not the full description (would require an extra HTTP call per job). The dashboard shows just the title/location/link for those.