Skip to main content
@deepidv/server is the official backend-first TypeScript SDK for the deepidv identity verification API. It wraps every REST endpoint in a typed, autocompleting client so you can create hosted verification sessions, scan documents, compare faces, run full identity verification, and screen individuals — without hand-writing HTTP, auth, retries, or file uploads.
This is the server SDK — it holds your API key and is meant to run in a trusted backend (Node.js, Deno, Bun, or an edge runtime). Never ship it to a browser or mobile client.

​
Why use the SDK

  • Typed end to end. Every input and result is a TypeScript type derived from a Zod schema — your editor autocompletes fields and the compiler catches mistakes before runtime.
  • Thin client. The SDK validates inputs, manages x-api-key auth, retries transient failures, and orchestrates presigned file uploads. All verification logic runs server-side at api.deepidv.com.
  • Web-standards-first. It uses only native web APIs (fetch, AbortController, ReadableStream, Uint8Array), which is what lets it run across Node, Deno, Bun, and Cloudflare Workers.
  • Single dependency. The only production dependency is zod for runtime input validation.

​
Install

npm install @deepidv/server
The package ships dual ESM + CJS builds with bundled TypeScript declarations, so import and require both work with no extra configuration.

​
The DeepIDV client

The DeepIDV class is the single entry point. Construct it once with your API key and reuse it — the constructor is cheap and each instance is independent. 
import { DeepIDV } from '@deepidv/server';

const client = new DeepIDV({
  apiKey: process.env.DEEPIDV_API_KEY!,
});
See Authentication for API key setup and Configuration for all client options.

​
Namespace map

Methods are grouped by domain. Each namespace maps onto a part of the REST API:
NamespaceMethodsWhat it doesREST reference
client.sessionscreate, retrieve, list, updateStatusHosted verification sessionsSessions
client.documentscanDocument OCR / data extractionDocument Scan
client.facedetect, compare, estimateAgeFace detection, matching, age estimateFace
client.identityverifyOrchestrated document + face verificationIdentity Verify
client.screeningpepSanctions, adverseMedia, titleCheckSilent screening (PEP/sanctions, adverse media, title check)Silent Screening
client.asyncJobsgetPoll long-running async jobsGet Async Job
The client also exposes client.on(event, listener) for lifecycle events — see Async Jobs & Events.

​
Three service tiers

TierPatternMethods
SynchronousOne call, one result. Image in, structured data out.document.scan, face.detect, face.compare, face.estimateAge, screening.pepSanctions, screening.titleCheck
OrchestratedOne call, multiple operations coordinated server-side.identity.verify
Session-basedCreate a session, the user completes steps, you retrieve results.sessions.create, sessions.retrieve
AsyncKick off a job, poll for the result.screening.adverseMedia, asyncJobs.get

​
Supported runtimes

The SDK runs anywhere with native fetch:
FeatureNode.js 18+DenoBunCloudflare Workers
All API methodsYesYesYesYes
File path inputYesYesYesNo
Uint8Array / Buffer inputYesYesYesYes
ReadableStream inputYesYesYesYes
Base64 / data URL inputYesYesYesYes
ESM importYesYesYesYes
CJS requireYesN/AYesN/A
Node.js 18 is the minimum because it’s the first LTS with stable native fetch. On edge runtimes there is no filesystem, so pass a Uint8Array or ReadableStream instead of a file path — see Configuration for proxy, mTLS, and service-binding fetch setups.

​
Next steps

Quickstart

Install, initialize, and make your first call in two minutes.

Authentication

API key setup, env vars, redaction, and custom fetch.

Session Verification

The hosted flow: create a session, let the user verify, read results.

Server-to-Server

Build a custom pipeline from the document and face primitives.

​
Resources