USAJobs Federal Jobs Feed — Daily-Fresh, Incremental, Agent-Ready

Pull every open US federal government job straight from the official USAJobs.gov API — normalized, deduplicated, and delivered as clean JSON. Built for job boards, recruiters targeting cleared/federal talent, workforce researchers, and AI agents.

Pay only for results. No subscription. A run that returns nothing costs nothing.

Why this feed

• Incremental by default. Each scheduled run emits only the jobs that are new or changed since last time — no re-paying for the same listings every day. Flip it off to pull the full match set.
• Clean, flat schema. Salary min/max + rate, GS grade, agency, location, schedule, security-clearance level, telework eligibility, close date, and a direct apply URL — flattened from the API's nested payload into one row per job.
• Direct from the source. Official OPM/USAJobs API. No anti-bot breakage, no stale "ghost jobs."
• Agent-ready. Predictable JSON, stable field names, pay-per-result pricing — drop it straight into an MCP workflow or an AI agent.

Output fields

id, title, agency, department, location, remote, salaryMin, salaryMax, salaryRate, grade, scheduleType, positionType, securityClearance, teleworkEligible, publicationDate, applicationCloseDate, applyUrl, sourceUrl, scrapedAt

Inputs

Field	What it does
keyword, locationName, jobCategoryCode, organization, remote	Search filters
incremental	Only emit new/changed listings across runs (default on)
maxResults	Cost guard — number of jobs emitted per run (default 1,000)
resultsPerPage	API page size (max 500, leave default)
resetDedup	Clears dedup memory so the next run re-emits everything

API credentials are supplied by the Actor owner via the USAJOBS_API_KEY and USAJOBS_USER_AGENT environment variables, so buyers don't need their own. (A free key is available at https://developer.usajobs.gov/.)

Billing model

This Actor bills via Apify's automatic per-dataset-item ("Result") event — exactly one item written to the default dataset equals exactly one billable result. The code never calls Actor.charge(), so there is no double-billing, and a run that emits zero new/changed jobs writes nothing to the default dataset and therefore costs $0 (the no-results notice is written to the run log, OUTPUT, and a non-default run-reports dataset instead).

Console setup: set the pricing model to Pay per event → Charge for each dataset item, and do not add any custom charge event. This mirrors the austin-permits-api monetization model.

Schedule it

Set a daily schedule with incremental: true and you get a live, deduplicated feed of just the day's new and changed federal postings.