Adverse media screening API.

Your first screening

Use a bearer API key from your server. Create a screening, poll its resource, then read the findings and coverage.

1. 1Create a request
2. 2Poll until finished
3. 3Review the evidence

{
  "subject": {
    "firstName": "Alex",
    "lastName": "Example",
    "country": "US"
  },
  "searchDepth": "standard"
}

A new request returns 202 and a Location header. A retry with the same workspace key and normalized request returns 200, the current result state and Idempotent-Replay: true.

Send the details you know.

Only first and last name are required. Use only identifiers you are allowed to process. All subject fields below belong inside subject, except the two marked top-level.

Search text is trimmed. Blank optional fields are omitted, country is normalized to uppercase, and email to lowercase. Unknown properties and control characters in search fields are rejected. Omit optional fields instead of sending null.

Identity, role and coverage are separate.

{
  "id": "00000000-0000-4000-8000-000000000001",
  "object": "screening",
  "status": "completed",
  "outcome": "potential_match",
  "failureCode": null,
  "referenceId": null,
  "searchDepth": "standard",
  "createdAt": "2026-09-01T10:00:00.000Z",
  "completedAt": "2026-09-01T10:02:00.000Z",
  "expiresAt": "2026-09-08T10:00:00.000Z",
  "subject": {
    "firstName": "Alex",
    "lastName": "Example",
    "country": "US",
    "city": "Boston"
  },
  "findings": [
    {
      "id": "00000000-0000-4000-8000-000000000002",
      "category": "fraud",
      "eventStage": "charge",
      "isPerpetrator": true,
      "severity": "high",
      "title": "Reported fraud charge",
      "summary": "A fictional report describes a charge against Alex Example.",
      "eventDate": "2026-08-20",
      "identity": {
        "confidence": "medium",
        "matchedAttributes": [
          "name",
          "location"
        ],
        "conflictingAttributes": []
      },
      "adverseRelevance": "high",
      "sources": [
        {
          "title": "Fictional report",
          "url": "https://example.com/report",
          "domain": "example.com",
          "publishedAt": "2026-08-21",
          "excerpt": "Alex Example of Boston was charged with fraud. The charge has not been proven."
        }
      ]
    }
  ],
  "coverage": {
    "complete": true,
    "searchesCompleted": 4,
    "sourcesReviewed": 3
  },
  "limitations": [
    "Illustrative data only. Review identity and evidence before use."
  ]
}

coverage.complete is separate from outcome. A result with findings can still have partial coverage. Read limitations, and review every finding before use.

Poll a stable resource.

Poll the Location URL only while status is queued or searching. Honor Retry-After, use increasing delays and stop on completed, incomplete, failed or cancelled. The Node.js and Python examples include bounded polling.

expiresAt is the result's expiry time. After it, a read returns 410. Deletion or retention cleanup releases the idempotency key, so reusing it can create a new screening.

Know the cost before you send.

Read GET /usage first. credits.standardCost and credits.enhancedCost are authoritative: paid plans normally use 1 and 2 credits; Unlimited returns 0 for both.

Compare prepaid credit packs and cost per screening. The browser playground and API use the same signed-in workspace allowance.

The request limit is shared by all API keys in the workspace. A workspace can have at most five queued, retrying, or active screenings. Read RateLimit-Policy and RateLimit when present. Protective network limits can also return 429, without workspace limit headers. Honor Retry-After.

Cancellation or deletion returns reserved credits only if processing has not begun. Inspect usage after a cancellation; do not assume that every cancelled screening is free.

Five operations. One workspace.

Base URL: https://adversesearch.com/api/v1. Every operation requires a valid bearer API key for an active workspace.

Save the screening ID returned at creation. The public API does not have a list endpoint.

Retry the same operation safely.

Use one opaque idempotency key per logical screening, shared across retries with the same body. A different normalized body under the same key returns 409. Keep personal data out of keys.

Read error.code, error.message and error.requestId. Responses also include X-Request-ID. A terminal screening may have failureCode; this does not mean it should be recreated automatically.

Review screening data security and the privacy and retention policy before integrating personal data. For rollout questions, contact AdverseSearch.