Harden my payment webhook handling for real-world failure modes. Review the handler for my payment provider's events. Implement: respond quickly (acknowledge within a few seconds) and do heavy work asynchronously, since providers disable endpoints that time out; return accurate status codes (2xx only when the event was durably recorded, 5xx when processing failed so the provider retries, and never swallow errors into a 200); handle events arriving out of order (a subscription.updated arriving before the subscription.created; look up current state from the provider when order matters instead of applying events blindly); store raw events for replay and debugging; and add a periodic reconciliation job that pulls recent events from the provider's API and checks I processed each one, catching whatever webhooks missed entirely (downtime, misconfiguration, the week the endpoint was broken). Don't process payment-side effects (granting access, sending receipts) more than once per event; keep the idempotency work intact. Don't trust event payloads for amounts without matching them against my own records. Deliver: the hardened handler with response semantics, the async processing split, the event log, the reconciliation job, and verification: kill the app, send test events, restart, and confirm reconciliation picks up what was missed.