career-ops hits Greenhouse's public boards API directly. No HTML scraping, no rate-limit risk, no AI tokens consumed during the scan itself.

What it does

The greenhouse provider in providers/greenhouse.mjs makes HTTP requests to boards-api.greenhouse.io/v1/boards/{slug}/jobs and parses the JSON response into the career-ops listing format. The slug is the company's Greenhouse board identifier (e.g. 'notion' for Notion).

When to use it

Use Greenhouse scanning when your target companies use Greenhouse as their ATS. Most US tech companies in the 50–10,000-employee range do. templates/portals.example.yml ships with 70+ pre-configured Greenhouse slugs you can enable.

Example

In your portals.yml: `- provider: greenhouse\n  slug: notion\n  careers_url: https://job-boards.greenhouse.io/notion` — then `/career-ops scan` picks up all open Notion roles.

What the scanner does

The scanner reads your enabled Greenhouse slugs from portals.yml, calls the public boards API for each one, and adds new, title-matching roles to your pipeline with no scraping and zero LLM tokens spent on the fetch. A single node scan.mjs run handles your Greenhouse, Ashby, and Lever boards in one pass, then filters the results by your configured title keywords.

Terminal

node scan.mjs

Output

Portal Scan - 2026-07-06
Offers found: 58 total
Filtered by title: 9 relevant
Duplicates: 6 (already evaluated or in pipeline)
New added to pipeline.md: 3

  + Notion | AI Product Engineer | greenhouse
  + OpenAI | Applied Engineer    | greenhouse

How career-ops talks to Greenhouse

career-ops requests boards-api.greenhouse.io/v1/boards/{slug}/jobs and reads the jobs array from the JSON, taking each role's title and its absolute_url. That boards endpoint returns listing metadata — title, location, and department — but not the full description, so career-ops pulls the complete job text with a follow-up request per role during the evaluate step, or by appending ?content=true where the board supports it. The scan itself is a plain HTTPS GET against a public endpoint, which is why it stays inside rate limits and spends no model tokens. Re-running the scan is safe: every URL is deduplicated against your scan history, your evaluated applications, and the current pipeline, so a role you have already seen is never added twice.

Gotchas

Greenhouse-API listings only include the title, location, and department in the boards endpoint. Full JD text requires a follow-up request per listing, which career-ops handles automatically during the evaluate step.

ashby
lever

Greenhouse