I tend to use thin events in a pub/sub system. For example if an entity is created, I generally won't include the full entity details in the event, just the id. Any consumer can fetch the full entity from the db if they need it. Is this still the right approach for eventbridge? I do lose the ability of having really granular auditability since the events don't contain the entirety of what's going on. With the "infinite scale-ness" of aws it seems like optimizing with skinny events matters less

Fat events for me - sometimes the lambda needs a consistent read so it goes back to the DB - but if was gonna do an eventually consistent read anyway, using the data in the event should work just fine

ah true it does allow for more granular rules. I like that