Kafka event consumption will be implemented as a kie server extension, following mechanism described in bpmn section 19.4, so it can be disabled, in case Kafka is not needed, by setting system property "org.kie.kafka.server.ext.disabled" to true. The capability associated is "BPM-KAFKA". The extension name "Kafka".

Kafka Topic names will be resolved from bpmn signal/message using system configuration properties at kie-server level. If there is no mapping for a particular signal/message, it will be suscribed anyway (assuming that the topic and signal have the same name), since the consumer will be configured to not create topics automatically (allow.auto.create.topics will be set to false), hence there is not risk of "topic creation scalation" issue in case of misconfiguration.

Message format will follow Cloud event spec. Data field need to match the structure indicated for that particular signal/message in the process structure ref. If this optional information is not present, in case of not primitive json object, a Map of key/values will be used. We are not supporting typed json unmarshaller.