Can I vibe code Ravelin?

Build a real-time Fraud Risk Evaluation Microservice in Node.js/TypeScript using Express, Redis, and PostgreSQL.

1. DATA MODEL:
- Create schema for `orders`, `customers`, `payment_methods`, `devices`, `risk_evaluations`, and `rules`.
- Store order details (amount, currency, shipping/billing address, IP, browser fingerprint token).
- Store execution logs for each evaluated rule.

2. CORE FUNCTIONALITY:
- Endpoint `POST /v1/evaluate` accepting order data, customer ID, device fingerprint token, and payment details.
- Implement Velocity Checking using Redis keys with sliding windows (e.g., card attempts per IP in 10 mins, orders per customer ID in 1 hour).
- Implement Rule Engine executing configurable logic:
  * Flag if billing country != IP country (via GeoIP lookup service interface).
  * Flag if order amount > $1,000 and customer age < 7 days.
  * Reject if email domain is a known disposable provider list.
  * Calculate aggregate risk score (0-100) based on weighted rule failures.
- Respond in <100ms with action: ALLOW (score < 30), REVIEW (score 30-70), BLOCK (score > 70).
- Endpoint `POST /v1/webhooks/chargeback` to receive chargeback notices, updating customer risk status and adding items to a global local blacklist.
- Simple Admin UI in React showing pending REVIEW orders with 'Approve' / 'Reject' buttons.

3. FAILURE MODES & RESILIENCE:
- If Redis or external lookup services fail or time out (>80ms), default to ALLOW but log a high-priority warning (fail open to avoid dropping legitimate sales).
- Idempotency key handling on `/evaluate` to prevent double-scoring on retries.

4. OUT OF SCOPE:
- Machine learning model training pipelines.
- Direct payment gateway execution or 3DS orchestration.

$ each button prefixes agent-specific run instructions · build your own product, never copy proprietary code, trademarks or designs