relaylet / hands-on experiments / 01
Break a payment.
Follow every cent.
A webhook arrives twice. A refund arrives first. A balance goes negative. Try it, and see what the books should do.
Choose your experiment
Step through a story, or use the controls to improvise.
Explore payment confirmation, repeated notifications, cumulative refunds, negative balance recovery, disputes, unknown usage, invalid signatures, early refunds and deliberately broken idempotency. Enable JavaScript to choose a story.
Experiment 01 / 09
Your first $20
Take the controls
USD · simulatedRefunds & attacks
Refund totals describe all refunds so far, not an additional debit. Signature rejection is modeled, not cryptographically executed.
Usage & a missing reply
Fixed illustrative amounts. Resolve actions supply synthetic evidence; they do not reconcile a real provider bill.
What just happened
0 actionsBlocked actions stay visible. Expand an action to see its postings. All amounts are USD.
- Your story starts with an empty ledger.
Both sides of the books
Signed ledger movements, not bank balances
Customer available + customer held + payment clearing + usage = 0. Clearing offsets the customer credit. A zero sum is necessary; the separate money check also compares credit to confirmed principal minus clawbacks, holds and usage.
Play here. Verify in the backend.
This simulator explains the payment rules with integer cents and a simplified USD usage model. It does not execute real webhooks, HMAC verification, durable database transactions, concurrent delivery or provider reconciliation. Those require the Python payment laboratory. No result here establishes production readiness.
Explore relaylet’s backend evidence →