Runs a whole sourcing brief on the server instead of making the caller orchestrate dozens of searches and evidence polls. The run expands the requested regions into countries, searches them three at a time against the candidate pool, ranks the union, gathers evidence reports for a bounded shortlist, and splits the shortlist into delivered profiles and screening leads. Countries your exclusions would discard anyway are skipped before any search runs, and listed in coverage.countries_skipped.

Returns 202 with a brief_id immediately; poll GET /v1/briefs/{brief_id} for the result. Debits 25 credits once, at submission: about fifteen pool searches plus up to twenty-five evidence builds reused from the shared cache, priced below the sum of the equivalent search and evidence calls. Polling is free of monthly credits. A 400 is uncharged.

A brief may name up to six regions, more than one search request may carry, because it expands and batches its own geography. Every criteria and exclusion field is validated by the same rules POST /v1/search applies, and every ceiling in the run is server-controlled.

Public evidence can never establish commercial tenure, employment, or any closed-source work. A delivered profile means a stated location that resolved to a single country offset plus at least one inspected, candidate-authored in-window commit in a requested skill; everything else is a screening lead with a stated reason.

{
  "criteria": {
    "languages": [
      "Scala"
    ],
    "domains": [
      "payments"
    ]
  },
  "geography": {
    "regions": [
      "north-america",
      "latin-america"
    ],
    "excludeLocations": [
      "India"
    ]
  }
}

{
  "meta": {
    "quota": {
      "month": "2026-07",
      "cost": 1,
      "rollover": 340
    }
  }
}