If you want to create reusable library you should avoid reading the configuration from any specific location. While there may be some exception to this rule, in general reading the configuration from specific location (say environment variable or property file) will limit the reusability and the testability of your library.
I think the best approach is to separate your code in two. The library itself without any dependency to Spring. You could easily configure it using constructor injection, builders or other suitable pattern. For example if you want to set the URL to connect to, you can use the constructor:
public class ExampleClient { private final String url; public ExampleClient(String url) { this.url = url; } // the rest of the class }
As your library does not depend on Spring or any environment it is very easy to be tested. And increases the reusability as it does not make any assumption about the environment it will run in. Even if you plan to use it for single environment, one day the environment or the requirements may change.
Then you can create your own auto-configuration that will create ExampleClient instance and configure it with the appropriate URL. How it will get the URL depends on you specific requirements, but in most cases is best to use Externalized Configuration. It will give you a lot of flexibility.
