The new deployment pipeline is ready for testing. I've set up staging with the latest config.

Looks good. I'll run the integration test suite against staging after lunch.

Perfect. Make sure to test the webhook retry logic — that was the main change in this release.

I noticed the retry backoff caps at 5 minutes. Should we increase that to 30 for webhook failures? Some third-party APIs take longer to recover.

30 minutes makes sense for payment webhooks. I'll update the config and open a PR.

Agreed. Let's also add a dead letter queue for webhooks that fail after all retries. We can't lose payment events.

I can handle the DLQ implementation. Already have a pattern from the notification service we can reuse.