Skip to content
Asqary Docs

Quickstart

This takes about five minutes. You will create a test key, hash a customer identifier on your side, and score one transaction.

1. Get a test API key

Section titled “1. Get a test API key”

Sign in to the dashboard with an email link, create your organization, and copy the test API key (asq_test_…). Test keys use test data, which never mixes with live data.

Keep the key on your server. Never put it in a mobile app or a browser.

2. Create an identifier key

Section titled “2. Create an identifier key”

Asqary never receives raw customer identifiers. You hash them with a key only you hold:

Terminal window
openssl rand -hex 32

Store the result in your secrets manager. See Identifier hashing for the rules.

3. Score a transaction

Section titled “3. Score a transaction”

With the Node.js SDK, which hashes identifiers for you:

import { Asqary } from "@twalle/asqary";


const asqary = new Asqary(); // reads ASQARY_API_KEY and ASQARY_IDENTIFIER_KEY


const decision = await asqary.score({
  transactionId: "loan_1001",
  type: "loan_disbursement",
  amount: 5_000_000, // minor units: ₦50,000.00
  currency: "NGN",
  occurredAt: new Date(),
  customerId: "customer_456",
});


console.log(decision.score, decision.riskLevel, decision.recommendedAction);

Or with any HTTP client. Here $CUSTOMER_HASH is the HMAC-SHA256 hex of the customer ID:

Terminal window
curl -X POST https://api-staging.asqary.com/v1/score \
  -H "Authorization: Bearer $ASQARY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "transaction": {
      "id": "loan_1001",
      "type": "loan_disbursement",
      "amount": 5000000,
      "currency": "NGN",
      "occurred_at": "2026-09-27T10:15:00Z"
    },
    "customer": { "id_hash": "'"$CUSTOMER_HASH"'" }
  }'

4. Read the decision

Section titled “4. Read the decision”
{
  "decision_id": "dec_01a0e30da731725886aab649d88d2f04",
  "transaction_id": "loan_1001",
  "score": 74,
  "risk_level": "high",
  "confidence": 100,
  "recommended_action": "manual_review",
  "reasons": [
    {
      "code": "amount_anomaly",
      "contribution": 25,
      "message": "Amount is 10.0× the customer's typical amount"
    }
  ],
  "policy": { "id": "pol_…", "version": 1, "version_id": "plv_…" },
  "environment": "test"
}
  • recommended_action is one of approve, step_up, manual_review or block. What you do with it is your decision.
  • reasons lists the rules that raised the score and how many points each added.
  • Store decision_id with your transaction so you can look it up and report its outcome later.
  • Scoring the same transaction.id again returns the original decision, so retries are safe.

Next steps

Section titled “Next steps”