Shield — Fraud Detection
Protect your payments with a weighted risk-scoring engine, configurable thresholds, and allow/block lists.
2 min read
Shield evaluates the risk of each transaction in real time and acts according to your thresholds — allow, hold for review, or block. It is a deterministic, config-weighted rule engine (IP intelligence, velocity checks, card BIN risk, device fingerprinting, blocklist matching, sanctions screening) — not a machine-learning model.
Live only
Shield does not run in sandbox — it only evaluates live transactions. You cannot exercise it with test data; validate your thresholds carefully before enabling blocking in production.
How it works
- For each live payment, Shield computes a risk score (0–100) from weighted signals.
- The score maps to a band:
low(0–20),medium(21–50),high(51–80),critical(81–100). - Your thresholds (
review_threshold,block_threshold) turn the score into a decision. Transactions aboveblock_thresholdare blocked outright; abovereview_thresholdbut belowblock_thresholdare flagged for manual review in the Dashboard.
Konsole·Request Inspector
POST
/payments201420 msShield15:04:22Request — Headers
AuthorizationBearer pk_live_••••••••••••a91f
Content-Typeapplication/json
X-Idempotency-Keyidmp_3pQ9…7Zk
Request — Body
{
"amount": 750000,
"currency": "XAF",
"email": "new@example.com"
}Response — 201
{
"shield": { "score": 62, "band": "high", "decision": "review" }
}Orchestration
block_threshold80
review_threshold50
decisionreview
Get started with the quickstart, then configure your blocklist and understand the scoring.