Hiring API

Assess candidates for a role and get a score you can act on.

The Hiring API screens candidates. You register a role, send us a candidate, and get back a scored assessment.

If you are helping people practise rather than deciding who to hire, you want the Preparation API instead.

What you can build

Screen everyone who applies. Create an assessment per applicant, send each a link, and collect scored results as they finish.
Score a resume against a role before a human reads it.
Put an interview inside your own product, so the candidate never leaves it.

Four kinds of assessment

Type	What happens	Good for
`resume_screen`	We score the candidate's resume against the role	Cutting a large applicant pool down
`ai_interview`	An AI interviewer conducts a call and scores it	Deciding who a recruiter should speak to
`coding_test`	The candidate solves problems in a sandboxed editor, judged against hidden cases	Checking whether someone can actually build
`prompt_engineering`	The candidate directs an AI model through realistic tasks, scored on how well	Roles where working with AI is the job

coding_test and prompt_engineering each need a module enabled on the organisation. GET /v1/assessment-types lists only what yours can create, so read it rather than hardcoding the four.

You do not manage interview rounds. You give us a role and a type, and we build the rest.

How an integration fits together

Register a role once, with POST /v1/jobs.
Create an assessment per candidate, with POST /v1/assessments.
Send them in, either with a link from POST /v1/assessments/{id}/launch-link or by embedding the session in your own page.
Optionally set a time with POST /v1/assessments/{id}/schedule.
Receive the result on your webhook endpoint.

Before you start

Keys start with ehs_. Get one from Settings → Developers in your Expert Hire dashboard. You will see the secret once.
Start with a sandbox key. It costs nothing, records nothing, and its assessments never reach your recruiters' pipeline.
Base URL is https://hiring-api.experthire.cloud.

A ehp_ key from the Preparation API will not work here, and vice versa. The prefixes are namespaced so a misconfigured integration fails loudly instead of producing results that look right and mean nothing.

Next: the quickstart runs one assessment end to end.

Reading your own account

GET /v1/usage reports credits and which feature modules are on.
GET /v1/analytics/funnel counts assessments by stage, optionally per role and date range. Read the field notes before reconciling: conducted is a set, not status = completed, because shortlisting overwrites the status in place.
GET /v1/members lists the humans in the organisation, so you can attribute work in your own UI.
GET and PATCH /v1/branding read and set the brand name candidates see.
/v1/api-keys lets you rotate credentials without a browser. A key can only mint another key in its own environment, so a test key can never produce a live one.