The reason was to simplify client initialization, because of assumption, that the connector will be used in seprate process or even in sharded mode, so unlikely will listen multiple buckets
It is still possible and in version 1.2.0 I will add corresponding constructor to override Couchbase and Kafka credentials. But to really share resources with java environment (which like kafka environment object inherits from core environment), you have to manually transfer pools and schedulers. We do not have easy way to clone only core part between environments.