Ashby Job Board Monitor

Monitor job postings from companies using Ashby (a popular applicant tracking system). Get clean, structured job data, refreshed on every run.

What it does

• Fetches job boards from Ashby-powered company career pages via their public API
• Normalizes postings into clean, consistent fields
• Filters by keywords, location, remote-only
• Change detection: with onlyNew, the run filters out jobs returned by previous runs
• No login required, no browser needed — pure HTTP, fast and cheap

Input

Field	Type	Description
companies	string[]	Ashby board slugs or URLs. E.g. ["stripe", "https://jobs.ashbyhq.com/notion"]
keywords	string	Comma-separated keywords to filter (e.g. "engineer, designer")
remoteOnly	boolean	Only remote positions (default: false)
includeCompensation	boolean	Fetch compensation data (default: true)
onlyNew	boolean	Only jobs not seen before (default: false)

Don't know the slug? See companies.txt for a curated starter list of 100+ verified Ashby companies.

Output

Each job is a clean dict:

{
  "company": "stripe",
  "job_id": "abc-123",
  "title": "Senior Backend Engineer",
  "department": "Engineering",
  "team": "Platform",
  "location": "San Francisco, CA",
  "remote_flag": false,
  "employment_type": "FullTime",
  "compensation": "$160k - $200k",
  "published_date": "2026-10-01T12:00:00Z",
  "job_url": "https://jobs.ashbyhq.com/stripe/abc-123",
  "apply_url": "https://jobs.ashbyhq.com/stripe/abc-123/application",
  "description": "Build things...",
  "scraped_at": "2026-10-12T00:00:00Z"
}

Pricing

Pay per event:

• Company checked: $0.003 per board fetched

$0.003 per company checked, including boards that return an error or 404.

• Job returned: $0.75 per 1,000 jobs

Run start: $0.00005 per run.

How it works

1. Fetch (ashby.py): Concurrent HTTP requests to Ashby's public posting API with retry/backoff
2. Normalize (normalize.py): Clean field mapping, HTML stripping, defensive parsing
3. Filter (main.py): Keyword and remote-only filters
4. Detect changes (state.py): The named ashby-monitor-state key-value store tracks seen job IDs and the last check per company slug. Filters are applied before recording seen jobs. onlyNew returns previously unseen IDs; changed and removed IDs are tracked internally without changing the dataset fields.

onlyNew state is shared per company across all your runs of this actor. A manual onlyNew run marks jobs as seen for your scheduled runs, and two onlyNew tasks with overlapping filters on the same company will split new jobs between them. 5. Bill (billing.py): Pay-per-event charges, respects buyer's spending cap

For developers

# Local test
python -m my_actor.main --local test-input.json

# Run unit tests
python -m unittest discover -s tests -v

The regression suite runs offline against a recorded Ashby board and simulated board changes. It includes fresh runs sharing one store, retry billing, and corrupt-state recovery.