Thank you for the comment, I have applied some minor changes for points 1 & 2. In case of RabbitMQ's HTTP endpoint I am very happy with the fact that it allows write operations that can for example change topology (e.g. add/remove bindings), what might come in very useful for debugging/incidents. And/or investigations, like listing all queues/bindings/exchanges. So it's more like administrative interface (and sometimes you need to use it, if e.g. our application needs to setup users/vhosts on their own).