# Feed field reference

Every field in the OpenTrain XML, JSON, CSV and RSS job files, with the unique key, translated copies, closed jobs and refresh rules.

Source: https://www.opentrain.ai/docs/developers/job-feeds/feed-fields/

OpenTrain publishes every open job in four bulk files. None needs a key or a sign-up. This page lists the fields in each one. For which file to pick, see [Job feeds and API](https://www.opentrain.ai/docs/developers/job-feeds/overview/).

| File | Link | Jobs |
| --- | --- | --- |
| XML | `https://app.opentrain.ai/feeds/jobs` | Every open job, plus a translated copy for some jobs |
| JSON | `https://app.opentrain.ai/feeds/jobs.json` | Every open job, once each |
| CSV | `https://app.opentrain.ai/feeds/jobs.csv` | Every open job, once each, same columns as the JSON file |
| RSS 2.0 or Atom | `https://app.opentrain.ai/feeds/jobs/rss` | The newest 100 jobs |

## Rules for every file

- **Refresh.** The files update by themselves through the day. Import a file every 10 minutes at most; more often brings nothing new.
- **Repeat requests.** The XML, JSON and CSV files answer `304 Not Modified` when you send the `ETag` you got back as `If-None-Match`, or the `Last-Modified` time as `If-Modified-Since`.
- **Closed jobs.** A job that closes leaves the file. Delete jobs that are no longer in the file. The [changes feed](https://www.opentrain.ai/docs/developers/job-feeds/api/#changes-feed) also lists each closed job.
- **Personal data.** None. OpenTrain AI is the company on every job.
- **Apply link.** Send applicants to the apply link (`applyurl` or `applyUrl`) and keep it as it is.

## XML file

The XML file uses the common job-board layout: one `<source>` with `<publisher>`, `<publisherurl>` and `<lastBuildDate>`, then one `<job>` for each job.

| Field | What it holds |
| --- | --- |
| `id` | The job's unique key. Use it to match a job across imports. |
| `referencenumber` | The same id with an `opentrain-` prefix, for tools that ask for a reference number. |
| `title` | Job title. |
| `company` | Always `OpenTrain AI`. |
| `date` | When the job was posted. |
| `expirationdate`, `closingdate` | Rolling dates that move forward with every build. They are not the date a job closes. Close a job when it leaves the file. |
| `url` | The job page on www.opentrain.ai. |
| `applyurl` | The apply link. |
| `country` | Countries the job is open to, comma-separated, as names. |
| `remote` | `true` for remote work. |
| `jobtype` | For example `contract, parttime`. |
| `salary` | Pay as text, for example `$15 per hour`. Left out when the job lists no pay. |
| `category` | Category name, for example `Audio & Speech`. |
| `languages` | Language codes the work needs, comma-separated, for example `en, es`. |
| `description` | The job description as HTML, in a CDATA block. |

**Translated copies.** Some jobs also appear in a second language. A translated copy has the same id with a hyphen and a language code, for example `…x163332859756544000-es`, and its `url` points to the translated job page. To list each job once, skip ids that contain a hyphen.

### Example: WordPress with WP All Import

1. Choose `job` as the record element and job listings as the post type.
2. Set the unique identifier to `{id[1]}`.
3. Add the filter `id` does not contain `-`.
4. Turn on removing posts that are no longer in the file, for posts this import made.

Automatic imports from the link need WP All Import Pro. With the free plugin, save the file and upload it by hand.

## JSON file

The JSON file has `publisher`, `publisherUrl`, `generatedAt`, `count` and a `jobs` array. The CSV file has one row per job with the same fields as columns; list fields are joined with commas.

| Field | What it holds |
| --- | --- |
| `id` | The job's unique key. |
| `title` | Job title. |
| `company` | Always `OpenTrain AI`. |
| `url` | The job page on www.opentrain.ai. |
| `applyUrl` | The apply link. |
| `category` | Category slug, for example `audio-speech`. |
| `categoryLabel` | Category name, for example `Audio & Speech`. |
| `pay` | Pay as text, for example `$15 per hour`. |
| `payType` | `PAY_PER_HOUR`, `FIXED_PRICE` or `PAY_PER_LABEL`. |
| `jobType` | For example `contract, parttime`. |
| `remote` | `true` for remote work. |
| `location` | Countries the job is open to, as names. |
| `countries` | Two-letter country codes. |
| `languages` | Language codes the work needs. |
| `datePosted` | When the job was posted (ISO 8601, UTC). |
| `closingDate` | A rolling date, not the date the job closes. |
| `updatedAt` | When the job last changed. |
| `description` | The job description as plain text. |
| `descriptionHtml` | The same description as HTML (JSON file only). |

## RSS and Atom

`https://app.opentrain.ai/feeds/jobs/rss` lists the newest 100 jobs. Each item has `title`, `link` (the job page), `guid`, `pubDate` and `description` (HTML), plus `category` when the job has one.

- Add `?format=atom` for the same jobs in Atom format.
- Add `?language=` and a language code for jobs in one language, for example `?language=es`.
- Add `?category=` and a category slug for one category, for example `?category=audio-speech`. The [facets list](https://www.opentrain.ai/docs/developers/job-feeds/api/#facets) has every slug.

In Google Sheets, `=IMPORTFEED("https://app.opentrain.ai/feeds/jobs/rss")` in cell A1 loads the newest 100 jobs, one row each.

## Links carry a source tag

Each job page link and apply link carries `ch=` and `sub=` values that say which file it came from, for example `ch=feed_csv&sub=csv`. Keep them, so people who come from your site are counted correctly. The [JSON API](https://www.opentrain.ai/docs/developers/job-feeds/api/#tag-your-links) lets you set your own `sub=` value.

## Related pages

- [Job feeds and API](https://www.opentrain.ai/docs/developers/job-feeds/overview/)
- [Public jobs API](https://www.opentrain.ai/docs/developers/job-feeds/api/)
- [Fair use and limits](https://www.opentrain.ai/docs/developers/job-feeds/fair-use/)
- [Job feeds change log](https://www.opentrain.ai/docs/developers/job-feeds/changelog/)
